Subtitles for Adopting DevOps @ Scale Lessons learned at Hertz, Kaiser Permanente and lBM

  Pause sub
we'll be doing polling in just a little
bit but before we do let me let me
introduce you to other as I said earlier
I think one of our best panels and
hopefully best webinars of the year if
we can go to our bio slides are
introduction to our panel
I've already introduced myself so you
know me but I three gentlemen who I
actually met all three of these well I
knew sanjeev earlier but John dragons I
met last year at IBM's interconnect
event februari and we've stayed in the
less februari we've stayed in contact
ever since then I've had this notion of
doing of telling their story in a
webinar it only took us 10 months but
we're here so let me first introduce you
to John left friend year jon is the vice
president of global like the
architecture it hurts either their
brand-new worldwide headquarters in
estero a sterile florida which is just
about an hour from you John welcome and
thanks for being here thank you well
thank you and our next panelist is
rockin with Ramon rockin truth is out of
New York or it's still out of New York
correct negative enough and upset as a
California about an excuse me san jose
over in Silicon Valley bargain with his
with kaiser permanente de large large
healthcare provider and he's the
director for digital 18 api integration
delivery and our third but certainly not
our lease panel members of good friends
and divine sanjeev sharma CTO DevOps
Technical Evangelist distinguished
engineer at IBM and also author of
DevOps for dummies which second edition
was recently released sanjeev welcome
and thanks for joining us thanks I'll
and tanks are John and also for 4i in
joining this panel i think is going to
be very exciting the penalty how
absolutely so let as i said earlier i
wanted to start off with a couple of
polling questions gentlemen for our
audience to give you a better idea of
who's listening here today we had a
quite a bunt
try number of people register and
hopefully though are taking our poll so
first question is is cloud an integral
part of your DevOps strategy and many of
you will probably say yes but we want to
know is it public is a public cloud
virtual private or private cloud or a
little of both the hybrid cloud or
perhaps no is the answer
so please give us your answers i hope we
got all of the choices in there and
hopefully it's not too doesn't take too
long to get us your answer will close
this question off in five four three two
one and survey says always like richard
Dawson so I I think I as i expected i
don't know if folks in my panel agree
but almost sixty percent 58 well over
half are involved in hybrid is an
integral part of their DevOps strategy
and that's both on-premise and cloud ten
percent of pure public fifteen percent
of pure private and he won't maybe this
number is a little higher than we
expected 18-percent cloud is not part of
their DevOps strategy or perhaps maybe
they have no DevOps strategy at all we
can we can discuss that a little a
little bit further on and let's go to
our next polling question that and this
is an interesting question that i always
ask when I'm interviewing people around
the world about DevOps so what is the
approach to the DevOps transformation at
your organization is it top down meaning
that's executive its executive driven
this executive cover its executive lab
or is it bottom-up you know this was a
grassroots kind of thing that's being
done on a project-by-project basis and
it's not a skunkworks per se but there
really is no very senior level executive
that's kind of mandating
and again if you can give us your answer
and I realize not everyone may fit into
one of these two boxes
sometimes it's from the middle out right
going both top and bottom but if just
one of two will close this often 54321
and our survey says wow that's close but
it's I thought it would actually be
close to the other way
fifty-four percent a bottom-up forty-six
percent say cut down gentlemen is that a
surprise to you a year ago maybe not as
not as much of a surprise me now okay i
think they're not side keep driving
yeah yg what about you annoyed i think
it's it's not a surprise
what I would like to maybe we could have
two twisted this question a little bit
to make it more interesting is when did
you start letting people always end up
here where it is you know a combination
of top-down and bottom-up the question
you know would be did you start here and
have you ended up where while you might
have started at one end you ended up
we're having to do both because that
eventually letting the what what is
needed and that's the pattern we're
seeing over and over again with multiple
clients of ours is you need to do but
you might start up and one end you might
have initiated with one and what you
know you'll end up wanting to do what
needed to the board top-down and
bottom-up and you know what one of our
attendees Vincent happen if I
mispronounce your name I apologize
that's exactly what he had in the
question section which is it's really
both and it probably does take both to
have a successful DevOps transformation
at some point someone else from the
audience asked in the poll results show
the total number of responders and
unfortunately the way the system is set
up it doesn't but i will tell you we
have just under 200 people here hundred
seventy-six are in at this very second
and over 75 76 percent voted on this so
you could do the numbers and and kind of
see for yourself
if it doesn't hurt your head you know
doing a little arithmetic but anyway
that that's that question I believe that
we have one other question jewels
yes ok here's our last poll question
before we dive in guys
which of the technologies are you
employing in your digital transformation
which horrible so and we're only asking
you to select one we realize it may be
more than one but if you had to pick one
which one would it be
containers such as doctors etc
microservices hybrid cloud or none of
the above
and again if we can get your vote I
don't want to waste too much time
because our panel is to have a lot to
share with you so we'll we'll close
things off here in just give it another
five seconds five four three two one and
that by the way for anyone watching
along at home now it's about sixty-six
percent of our attendees which are up
over 118 our voting well thirty-one
percent of docker 31-percent hybrid
cloud and twenty-two percent none of the
above
I would suspect that many of the people
doing microservices are probably using
containers as well I'm Roger what it
what do you think I think that none of
the above is interesting where it's
probably indicated its probably
indicated of early adoption against or
maturity of where they are
there's some flavor of this that is
probably needed and what we are
generally calling DevOps it so it could
also mean this a lot of automation that
already exists and DevOps has been as
DevOps but they don't use some of these
technologies so it's then but none of
the above isn't surprising to be that
high
yeah agreed ok quiet
are they said you got em I'm a jump in
now our I think there's also the from
one of the wall question and you know
that this question can be misleading in
a way what does this transformation will
right because if a customer has if I
company has already you know heavily
digital heavily disguised and they have
automation in place
do they put under none of the above are
they in or do they go under hybrid cloud
cutting that definition of this
transformation becomes a it can be
subjective you know i would like to ask
this question again I keep twisting your
questions and I apologize is an army of
what technologies are you are you
adopting two to talk about innovation
try to uh to be more innovative writing
to be on that cutting edge and
experiment with new technologies or new
business models are you using any of
these technologies for that I think that
would have might have reduced that none
of the above
uh-uh you no answer that big number
there but that's again just my my
opinion on it so maybe people can post
some other comments in there in the chat
session I you know we are regarding
innovation maybe we can now get some get
some conversation with running that
absolutely
with that being said let's by the way
folks were sharing these results on
Twitter for anyone following along our
Twitter head twitter handle is a DevOps
dot-com that's the evo PSD OTC om I that
being said let's jump and it's in our
chat window let's jump to our slides if
we can
ok so I've introduce you to our speakers
and I i believe John laughter anyway
from hertz is going to lead off when we
moved to first first slides there please
okay John let's take it away alright
hello everyone my name is Jonathan air
is Alan mentioned um the head of
architecture for hurts and hurts been
around for a long time all of you have
heard the brand I'm sure we're nearly a
hundred year old company and the company
did not focus a lot on technology in the
previous 20 plus years actually and so
on arriving here with many of my closest
IT friends as part of a large rebuild if
you will or transformation of the
company we found that we had huge
technology debt is did a lot of effort
lot of time and effort assessing systems
and realize that we we weren't going to
be able to upgrade a lot so we needed to
make a major transformation moving to
cloud and and leveraging cloud and SAS
technologies was the logical course of
action and then as we kind of keep
pressing down that that that road it
appeared to us that what we need to do
to do was focus more on buying and
building less so there that's where the
phrase the config ops vs DevOps phrase
comes up and talked about this a little
bit at interconnected IBM last year and
that the focus early on for us is to buy
products by SAS offerings try to
configure those offerings within the
paradigms or within the rule sets that
are given to us by the various a
spider's and try to do as little custom
code as possible until we get further
down our transformation where we start
having to write or or build custom code
so but net are the hurts transformation
is oops I think I want two slides sorry
it hurts transformation is is one that
is rooted in a lot of Technology debt
and is aiming at a an endpoint type of a
strategy so because of that our journey
actually started very top-down very much
a CIO let effort if we're gonna do this
let's go to we're going to do it on the
cloud a lot of effort by myself and
others in establishing you know that
strategy the cloud for strategy
understanding where we might choose to
use different things like containers vs
microservices where we're using both
what clouds we buy in what services we
buy what large partners and firms do we
align ourselves with to start the
process of transforming our core
applications because for hurts this is
about our course these are core
applications as these are the the
applications that we rent cars with
these are the applications that we
manage our fleet with these are the the
applications that we track financials
with and you know all the core things
that it takes to run our business is
what we're focused on transforming to
the cloud as part of our are a
company-wide DevOps strategy so like I
said it was more of a top-down effort
here it's it's now getting you know
ground swell and mix well my guess you'd
say and and there's a lot of enthusiasm
around the transformation but it
definitely started you know top-down
also I you know the the comment there
about the strategy the culture of the
company was not like I say we had not
done a lot with our technology for more
than 20 plus years so the culture was
not ready for the amount of change the
speed of change the end the target
really the target state of changing
everything and and a lot of very
difficult discussions you know where an
entire platforms or are being sunsetted
and going to end being replaced by brand
new systems being either a purchased and
and configured or in some cases written
but complete complete the take out of
old systems and so the impact to the
culture both inside of IT and outside of
IT folks that have been around for a
long time
you know it's it's a tough pill to
swallow two to think of an application
that you may have worked on for 20 years
is basically be been deemed you know
that it's the best thing we can do with
it is is replace it holistically vs
evolving at any anymore but at the end
of the day it was good for the company
and an alliance to some of the other
transformations that were happening
around a leadership and things in the
company so you know that this strategy
and and the top-down aspect of it it
goes all the way from our board down we
miss this is a our transformation and
aspects of the technology transformation
i have been discussed openly you know as
part of our public disclosure of our
numbers for you know where or public
company so our investors were are
informed of all of this and and so this
is a very very public very a well-known
if you will
devops journey that that were embarking
on multiple years it'll be over four
years when it's all done got a little
over three left to go but very very
committed as a company that to doing it
so to kind of close out without going
into a you know a lot of detail could
talk for days on this stuff and
obviously you guys want to hear from the
other panelists as well so I I wanted to
bring out a couple of these challenges
the things that have come about in the
last year and a half of really doing
this at scale at speed
you know tens to hundreds of millions of
dollars of effort and investment so
really a hundred percent committed some
of the things that I've noticed that
might be able to help you a little bit
the first one that point cloud isn't
magic that's a little bit of a note
gently poking towards our CIO and and
other you know leaders within IT and
outside of IT at
there's a big misperception that because
of the because the infrastructure
barriers so much lower when it comes to
cloud there's an assumption that the
barrier is that low for everything that
everything just is easier or
dramatically easier or doesn't even need
to be taken care of just because its
cloud and those of you that are doing
much within cloud I think he realized
that's not the case but this is a
misnomer I think that especially if you
do much in the way of financial planning
within your company and things like that
you gotta make sure that the that this
belief is not purposive among surely
your leadership because once they get
the blue beef isn't that cloud is Magic
then then the assumption is that it is
always cheaper in every way shape or
form and though there are cost
efficiencies of a we know with with
cloud and it is definitely the UH the
best way to move strategically in in our
current I tea set
it doesn't necessarily mean that every
everything you do there's going to be
cheaper like you said the resources
associated with cloud still need to be
fully understood
it's a different resource model and you
know if you're if you end up in a pure
hybrid mode where you're both data
center and cloud again that's another
area where this looking at your cost and
understanding you know what the cost
structure of those multiple environments
is and what resources it takes and and
to run it r is very important and it's
very important to make sure that
leadership throughout the organization
understands that so that you're not
caught trying to support things in a
cloud environment without the proper
staffing my second point is a point
that's a counter to some of the large
cloud providers actually who I you know
don't have a problem with you being in a
favorite state where you're using some
of their real estate and and some of
your own for hurts this would be a death
sentence if we were to take this
strategy on the reason being is we have
a legacy of not being able not finishing
projects properly you know shutting down
the old old application and things like
that so you end up with a lot of
duplication and and stuff and if we were
to head towards a hybrid environment for
cloud that hurts we would end up in the
same same place we would have died I
would spin up you know cloud instances
and i would still have my data center
stuff and I I wouldn't really get to
some of the financial savings that we
want to achieve as well as the
operational savings and the and the uh
higher-up times and all the other
advantages associated with with cloud
but you know so much of your other data
center if you think about your data
center so much of the cost of your
datacenter gets tight tied up in what I
call you know the last week shut down
the data center in other words the
network has to sit there till the very
end your storage has to sit there till
the very end your security has to be
there till the very end all of the
communications and connectivity for when
a lot of your highest dollar items are
the last pieces to be removed if you
will from a data center so you know if
the strategy is to move to cloud for
cost reduction or datacenter reduction
you gotta go all the way or it's just
it's going to be a failed effort if
you're using the cloud to extend your
data centers and extend services and
applications and make things more
flexible the if the justifications are
different than then hybrid maybe a great
place to go but if you're trying to get
out of the data center business hybrid
is a death sentence that's absolutely
so and John you know them I'm really
glad you ended with that because to me
that was one of the unique things I
remember watching you up on the main
stage unless Vegas last februari and
thinking to myself wow you know just wow
they are they're taking a bold bold plan
of of transformation here where I think
many people kind of stick one toe in the
water than stick three toes in the water
and maybe put their leg in by doing this
hybrid kind of two step if you will or
maybe it's a four-step or more where you
guys are really saying hey i'm at the
data center now but as we turn on our
our public cloud you know infrastructure
or you know where we're shutting it
would not you know what you're shutting
down what we have back at the on prep
and and kind of skipping that hole deep
frogging the whole hi grid I step if you
will and my experiences is that that's
not that's not what many people the
majority of people have done and I i
wonder i mean obviously you guys are
committed to it and you already well
into it and it seems to be working but
were stymied people say hey you know
this is not the way that the people try
to discourage you from doing that well I
think the key word you said it was is
committed and it and both connotations
are are appropriate
I mean there have been multiple times
many different either suppliers or
industry folks I do a lot of work with
IBM obviously and and folks have looked
at at the plan the big plan multi-year
plan and looked at all the different
major initiatives that we are going to
have going on in pair LOL and yea
absolutely people think we're crazy you
know the
and and it's it's is nothing I've ever
done at this it to this extreme but um
it you know i don't want to spend too
much time talking about we know that the
state of affairs when we when when I
arrived here but a lot of it has to do
with you know where we were and the only
way to get the company where it wanted
to be to enable the business
specifically marketing sales and revenue
management that which are you know key
aspects to the rental car business and
IT was holding back those portions of
the company in dramatic fashion
especially marketing and there's some
very basic functionality that some
things that they should be able to do
that we simply can't up we can't do
because of our back-end systems and in
some cases those systems are in excess
of 30 years old and they're still major
major major cogs and you know in the
machine that makes make us rent cars and
so you know II you start getting a
system where you're only as good as your
weakest link and then when you start
looking at DevOps and you look at the
web and mobile and how fast that's
moving and the expectation for people to
be able to do anything they want you
know from those devices through those
mechanisms if your if your back-end
can't support it there's just simply not
there's not a lot you can do you can
make the front-end look as pretty as
nice as you want but if you can't
support the function of the box that
somebody's clicking on then you simply
can't move forward and that's that's the
situation we found ourselves in so
that's where we step back and just said
we
it's all gotta go just got other
rebuild it and if we then we looked at
different timelines and scenarios and if
we did it over five or six years you
would just stranded too many pieces
along the way we just said we gotta rip
the band wait and we've got to do it all
at once we gotta do it fast and with
commitment
yep you know what John we have some
questions already from the audience
there will come back to at the
conclusion i have some more questions as
well but in the interest of giving
everyone their fair share
let's move over to argue and and I
understand a little bit about what he's
done at kaiser regular you want to take
it from there next slide
yes track next slide looks like it's
working
there we go okay
hi everybody my name is lagu ramen i
manage an engineering team here at
kaiser bitch is in api's and
integrations and majority of the api's
that we been endeavor are between our
consumer facing our member facing or
public-facing oh my team is currently
not doing API integrations that our
workforce or providers and doctors and
so on
we are very public focused and what we
see what we see is we bridge the gap
between systems of engagement and
systems of record so that's a context of
what we are doing a quick background of
how geyser is and what is how what we do
we have about ten point six million
members hospitals large number of
medical officers physicians nurses on
this is some data from the previous
quarter that we have and some some
notion of what happens under bitch to
the channel this is this data is a
little old and it's probably doubled by
now in terms of a usage that is there
this gives you a context of what is
happening in terms of number of people
how many results of days
this is important for us because we it
matters
for health patient safety people's lives
are catching on a daily basis for when
you want to schedule an oil change
it's okay if you don't have an
appointment today and it works out on
the weekend and you can schedule it
around that you have a problem with
coffee and a child you need to it right
away and we see more and more people
adopting the digital channel to get in
touch with their doctors two are meant
to do whatever this and like being the
most important thing and healthy very
critical it is to change the way health
care is experienced and delivered and
enable digitally so that's a context of
what this go to the next slide and I
think just before i go there is number
of topics that John touched on but
there's a lot of similarity between
what's happening at kaiser and herbs and
it's probably true in some of the larger
enterprises and I'll come to that before
we going into y DevOps of water is
driving the need for or why did we just
do it and it's possible that a a lot of
companies and not a first are doing
DevOps even before the term happened we
can say enable the best automation we
can label it as tools and technology
pipelines and saying be the white white
is new word of DevOps what's different
in this and so on so we'll come to that
but i think they're in from kaiser
standpoint there are three or four
things that are key we had regulation
changes that happened a few years ago
that changed the way the markets operate
we have digital disruption what I'm
talking about is the advent of a
separate frameworks whether the
JavaScript frameworks and tools or
technology or the ability of
provisioning systems in the cloud our
containers are the amount of tools and
technology that is available IP
landscape is significantly changing
the consumer behavior is critical that
is now a very clear need by a consumer
about how do you want to do thanks
and I was just listening to john and
said hey hurts has a challenge and
everybody knows about it with uber and
lyft answer on so what is it that why
would I apart from the business model
they need to be able to access whatever
they want
it's really a squeaky as and
yes you can paint or and have a
beautiful friend day but the backend
systems don't change is a different
problem and I understand Giants and hurt
this philosophy how they're going to
change the back ends and so on we don't
have that luxury or we don't have that
leeway of changing some of the backend
systems we are think of kaiser as seven
companies within one branding which is
seven regions and they all have their
regional systems their original
membership and regional health care
pharmacy applications and so on and over
75 70 plus your period these systems in
some ways are antique and may not be
updated so how do you expose digital
functionality
how do we react to the competition to
the changing tools because things that
we need to do in the Bunty speed were
where certain systems with a different
pace than others
our objectives are we have to know what
the cast we have to be able to market
faster faster time-to-market we have to
be able two reels be neat better
productivity and AP is our prime
candidate that I'd bridge some of this
but it is multiple pieces that need to
be touched
along the way I want to go to a couple
of topics that we I think we've spoken
about and saying is dave arts and
culture or is it an enterprise
what is that really need i think the
word culture is used a multiple ways i'd
rather take a step back and save an
enterprise has your period of time
I certain things we have had transpired
to the organization where you say okay
somebody's going to be managed on
infrastructure in the beta centers and
therefore its infrastructure group and
somebody's going to have to scale our
critical massive which you have skills
in development and then you made the
engineering department and so on these
signs started farming the word sino is
use i'm using it usually but it is how
organizational structure created these
organizational boundaries not from a
functional standpoint DevOps is really
talking about hotter and one of the
challenges that any large enterprises
going to face is putting an assembly
line that traverse through an
organization bounty structure tunes are
easy to do but if you have control over
all of the tools from end-to-end you
have much better control but the
cultural challenges are how do you after
process or implement a pipeline and
implement automation that spans across
functional units that were created a
point in time to certain needs
it's not and that's the barriers that we
find and say each organization functions
in a particular way because they were
built up in a particular way to solve
specific purposes of particular point in
time so you could have an organization
that say oh I'm going to have my data
center operations outsourced to better x
and every change request goes and once a
week on Saturday and that's the process
it that's the st that's been set up if
the 15 what's startling code if you
hours
well you have classes that are set in
place you need to change the mindset and
understand ok this is what we need what
its needs to change so just that's the
country bad in terms of saying you're
trying to impose or you're trying to
adapt a process that goes a class
functionally
set principles are functionally set
paradigms that exists because of
appointment time what a box does is
helps put that tooling and customer
provider relationship in each of these
organizations that say if you can swim
that are then you have a way to do think
stopped suddenly becoming evident the
same development team never bothered
about doing badly other about how this
has told was executed our how many JDM
and so what was their production army
casters come to that because we want a
piece of code astronaut environment and
you shift it over and somebody was there
somebody else was responsible for
everything that pops up saying you have
no visibility into and say I need to
understand how this piece of code there
it puts the onus of load testing
capacity planning security and turn
right from the Emperor to the tester to
the office that all in the game so
that's something that pops has to have
to foster and saying we had all of
distributed i think one of our leaders
had said earlier we're all in this boat
it doesn't make a difference of which
side the wholeness you can say I'm going
well on my side the whole is that on any
site that would be all good up there
officers something similar where you
need buy in to say we need to have a way
that we all have visibility and cleared
handoffs and have automation couple of
things that being and I think just
reacting to what some of the pole
answers and questions were cloud and I
said I saw a higher number of other
people said they were not using cloud
that's not going on there it's like a
great i think one of the reasons why
Isis we flower is PR new because of the
ability and the numberless that we need
to be able to scrape infrastructure a
few years ago always meant black against
a friend
since the environmental well today if
say that if that is a minimal as a
service and if it is a cloud we're happy
eyes to be able to provision BM sent
through something that is something that
is allows its mark time to market and
how quickly can I do it so it's
definitely need based where you see if
infrastructure is not part of your
challenges than know if it is
yes some of the challenges that we
really have from or I'm going to go back
to seeing one of the challenges that we
have we have multiple assistance
multiple regions and need for my
releases to go simultaneously multiple
consumers of the API so we have x apps
at least a two year for fishing
absolutely absolutely geyser we have the
idea channel keep it up august the
channel and we have certain call centers
calling me p is to access data that a
member has been see all of these are
added cadences how do you feel something
that can be nimble and move quickly and
releases which we means-testing capacity
would have not testing sightings and
tradition cycles we want to be in a way
to say okay if that is the consumer and
the definition of consumer at this point
for B is an application if they want to
go over the different version of an act
i want to be able to provide a service
status only consumed by there and I want
to be able to apply no jvm or a
container that is going to be continued
better i don't want to spend three
months cycle or Hardware can I provision
it in a way that is on demand that's
elastic and isolated from everybody else
how do i have isolation and scale and
get the market so those are some of the
drivers that Allah our state
how can I do frequent releases and the
infrastructure and cloud that's where
some of these patterns fun and play
which is saying that it's my combination
of Michael services
it's a combination of containers it's a
combination of cloud specific needs
liberia doing that I want to touch upon
something that John said earlier cloud
is not magic thing I think it isn't at
kaiser we recognize it anywhere
what I think that people are and
specifically it needs to be understood
as to what costs are Catholics what cost
of objects
how do you do that and what is the cost
of studying what's the cost of running
it what's the cost of monitoring and
John Berryman said it's not magic it
comes with a price and ease-of-use
should not be mistaken for lower-cost
ease-of-use helps that certain things
but you need to understand what it costs
to do that there was another item i
think i want to touch on a lot of the
tools that we have start-ups at other
places where they are able to hold a
very nimble be quickly i think as
enterprises we may not be able to both
nimble because different investments
that have been made righty different
structures that exist organization but
being able to adapt and adopt certain
news that makes sense for reading your
objective be selective about what
problem your song baby and then pick
those tools in your pops generally I
think that's how I say choose 13 figure
out the goal that you're trying to do
and then start going up
I don't that the last comment i want to
end up is I think the bottom-up vs talk
down privilege i think it was somewhere
I think a lot of places are so going to
be reading out the sponsorship and the
tribal could be somebody else I think
that are in a large organization like
Kaiser there are units and groups that
are doing pops in so-called a way to
optimize and making big
past and automate things and so on
here we have the pipeline into end
perhaps know how do we stitch all these
pipelines and have a chain that goes
across each of these kind of ups i think
that's something that is important and I
think kaiser we have taken a very
conscious decision to say we are going
to invest in this to automate and have
more environments and more automation to
end even faster steadily and this is
executed sponsorship that allows us to
do this so that some notion of something
that was already happening and then it's
decided to investigate better so just
kind of middle of upward and that it's
us they're saying we all need it
I did something and wait for questions
you know where quick we've got to give
some trivia chance but we have a lot of
questions and we're running love but our
time but Reggie let's turn it over to
you to bring this home and get us over
to the question section afterwards are
ya so let me kind of far more people
have access to the slides and you know
this has been a great conversation you
know a set of presentations so far from
Jon & goo
so my uh you know my role is a bit
different in the sense that i do not
work while i work at IBM i do not work
in IBM development so i am not the
person helping our developers here at
IBM adopt tell us we have our entire
team dedicated to that and we have
probably one of the largest global
companies you know which is focused on
an order opting DevOps my experience and
where I think I have a unique you know
seat at the you know seated in the
stadium of the office adoption is that I
get to observe and work with hundreds of
customers you know including your
sanitizer around the world to help them
adopt offs understand the challenges
they are facing and along the way learn
a few things and one of the things we
have been doing that at IBM between me
and the team i work with is looking for
patterns and anti-patterns right and you
and some of them uh you know you would
be it will be mentioned by john and
regular by the slide you're seeing here
is our attempt at modeling some of these
patterns right and I alluded to this
earlier on right about you know
definition of innovation right and
definition of this transformation and if
you look at you know what Raghu and and
John spoke about as to what the business
drivers were of the transformation of
their organizations when true they did a
little bit of what did it like that
hurts and you know I was you know I
don't we don't have too much time to go
back and forth but i'd love to hear you
know maybe after after I've gone from a
couple of my slides here Jones comment
on this
it sounded more like an optimization
efforts to modernize what you already
had rather than just creating brand new
business models are brand-new uh you
know engagement systems of engagement to
work with customers
it was you know your core business
systems which you were modernized right
in this model you see here on the screen
we are showing that there that all
applications and our DevOps adoption
efforts are spread across the spectrum
at one of the spectrum which of the
bottom of this slide you have what is
known as the industrialized core these
are when you are modernizing are
optimizing your core business
applications right you you already are
running a business you already have IT
systems which are running that business
and you're focusing on improving them
either totally more organizing them are
just optimizing what you have in place
the problem is well understood the
solution is well understood you might be
using new technology to deliver it at
the innovation at which is the other end
of the spectrum and please not only work
by mortal here right i'm talking about a
spectrum and at the end of the day you
wanted to be a normal distribution
across these two ends and the other end
of the spectrum you have what we are
referring to as the innovation edge here
is where you're focused on creating new
applications are new capability
T's in your applications to really
innovate to come up with new business
models to call the new ways to engage
with your customers you might be
explaining a lot of experiments here to
really not just understand the solution
but also in a lot of cases to understand
the problem and find the problem
solution match so how you adopt DevOps
varies based on better for the kind of
applications you are adopting Devils for
is on one end of the spectrum or the
other or somewhere in the middle
citing this is where you know we've had
a lot of success in helping customers
fine-tune develops adoption and not take
missteps are not fall into some of the
anti-patterns know what is interesting
here and what you see in those red
circles and lines in this in this chart
is that these are not isolated right if
you're talking to a large company with
with multiple applications there are
always dependencies across these
multiple applications so you might have
these multiple delivery pipe lines when
you have focused on eliminating the
silos as who was talking about which are
between the different functional areas
right you eliminated or reduced the silo
between seven knots right that's what
they've started or between Devon qar
between the business and IT side are
between your operations team which is
gathering feedback and running the
systems and your delivery routine which
are building and delivering the
applications but you also need to
eliminate the swim lanes for lack of a
better term which is created between
these delivery pipe lines because if
there is dependencies if you do not
optimize your delivery pipeline across
the spectrum you can only go as fast as
your slowest delivery pipeline if you
have a dependency on it right so it your
unoptimized areas can create a lot of
drag when it comes to wanting to go
faster wanting to innovate also if you
look at large IT organizations one other
common pattern we're seeing is that most
of the likely budgets are being consumed
by the industrialized core by running
maintained by keeping the lights on
optimized
I think those industrialized core
applications frees up resources to
invest more in innovation so they have
to go hand-in-hand
so there's a model we've been working a
lot on if you come to you know
interconnect will be talking a lot about
this as we go forward and let's move to
the next slide I don't know why have i
have control here so we can go to the
next slide jewels that would be great
excellent so one other thing is we have
done at IBM is we've started a capturing
our learning lessons what we're adopting
the officer scale across IBM but also to
the hundreds of customers we work with
and we have captured this always
something we're effort where the IBM
garage method which is a methodology we
are working on energy you know it is
something we are continuously improving
on as we learn more and more lesser than
as the experiment by frankie also with
all the content but something we
released earlier on which I had the I've
actually had the privilege of a
co-leading their foot was the divorce
reference architecture which is what
you're seeing here on this screen and
this is a available free of cost to
anybody anyone of anybody who wants to
go download it and look at it and
understand the practices and the
capabilities behind it
I at this URL on the screen and it is a
platform and technology agnostic right
so you can implement it based on your
technology stack which you might be
using and it can be implemented both
whether you're doing innovation-driven
adoption or optimization driven adoption
could go to the next slide please
actually I do have control very good so
the well one thing we do use to engage
with our clients and if you are you an
IBM customer or want to be an IBM
customer and are wondering how does how
do we engage with customers with both of
our clients to understand what would be
the best acts for them to adopt DevOps
how do we balance that top-down and
bottom-up which we spoke about how do we
balance the optimization the innovation
a kind of applications how do we
optimize what is the right place for
close to be introduced and in what
manner how do we decide whether we
should go into microservices
our containers are not one of the things
we do at IBM is near on this day long
halfway along with other workshop called
the DevOps optimization innovation
workshop in this workshop our goal is to
identify variants where is the waste in
your delivery pipeline so we take an
example of pipeline and do a value
stream mapping of the pipeline looking
across awesome looking for areas where
they might be a bottleneck sources of
waste what you wanted to find them then
you can narrow it down and say which
develops capability can help you address
that area of west or a bottleneck as you
call it so it helps you create a first
pass on a skeleton of a privatized
develops adoption road map from which
you can move forward and going to a
large-scale develops transformation so
if you're more interested in this you
know we needed more than happy to to
follow up during the Q&A session we have
several large customers of course you
already heard from kaiser and-and-and
hurts today but we are also several
other IBM customers across the world
different sizes here are and different
geographies we have helped adopt we've
already spoken a lot about you know our
are interconnected coming up in 2017 but
we also have some interesting videos
which were published from past
interconnect so that's what's on this
slide here but with it i'll pass it back
to Alan so we can take the few minutes
we have an open it up to any thank you
sent you i know we we we kind of put a
gun to your head on time there I
appreciate your moving along I don't
want to take a lot of time chatting
either accepted say if you can get to
interconnect that's right first or all
three of these guys and i highly highly
recommend that you if you can get there
go the links are there
let's jump into questions first this was
a question I think for John there was a
few questions on this subject
how do you manage the risk in terms of
data security didn't well you have a
very dedicated CISO that's a peer of
yours that works hand-in-hand with you
and
you know as we get into some of the
big SAS deals before we close those
deals out obviously we we bring security
to the table and make sure that we are
able to go meet the the to be security
parameters for the company i say the to
be because where were you know raising
our security profile as we go
you know into the new world so what has
been ok before is is no longer okay
going forwards or raising the bar and we
just said we we get ourselves to where
were at the threshold and getting ready
to get into detail statement of work and
and and signing up for large blocks of
work and and we bring security and and
validate the details of of the
particular security profile you know
needed for that for that particular
deployment what word and then what will
also do as we move into an environment
closer to like that of kaiser where you
know we have developers and people
starting to kind of see more traditional
DevOps spinning up their own things will
have profiles and templates that you
know meet minimum standards already in a
built-in provision so that folks can can
use those services and and we don't have
to worry about them breaking security
rules
excellent
I the the next question is again was
around in directed to you John around
info second public cloud it's one of the
major impediments that this person's own
organization that's forcing them to stay
hybrid was there a specific public cloud
or yo that you more than you very get on
well um we're not i've got i will I
struggle with the term hybrid I am I am
I consider myself hybrid but I'm not
hybrid as to what most definitions are I
i have no cloud environments in my own
data center
I'm hybrid from the perspective of i'm
in multiple clouds and i'm also in
different clouds at different levels so
yes we are doing some public cloud yes
there are some info SEC security
concerns that you need to have with all
the different clouds depending upon what
it is that you're trying to do and and
everything but that's also why virtual
private and private spaces available to
all of them so I've got a large
investment in IBM private a blue
mixcloud i have environments with Azure
I've environment and then I have a large
SAS environment that's being built out
through a partner in AWS that is
actually a dedicated environment that
I'm the they're dedicated environment
within a day of AWS we will be doing
some AWS public and so basically the
there's no there is no perfect answer
each every project that you have
depending upon those infosec
requirements may or may not inhibit you
from one cloud one public cloud or
another or cause you to take on
additional things you know I my previous
destination before coming to hertz was
at gmac financial or ally financial and
we were able to do some things move some
things for that bank and financial
institution that people didn't think
were you know that would that we could
do before and it's just a matter of
asking the right questions and then and
sit down with security people trying to
work out what you know what truly are
requirements vs maybe best practice and
and how do you get as much of that
profile knocked out as possible
absolutely ok another one for you John
was what was the legacy legacy
application architecture it hurts that's
being replaced
I have lots of microfocus cobol you so
you want to start if you want to start
with mainframe cobol and could take go
all the way up to the very early stages
of Sola and then stop and take 15 years
off thats
that's pretty much the hertz legacy
environments not quite that bad but it's
very old
got it uh let's jump to a kaiser
question here
Andrew this is for you disguise a user
public cloud for all of its systems or
private cloud how did they overcome
again how did they become security
challenges of having sensitive data and
cloud I think it's not going to be
private
we are not given the snow needs and I
don't thi we are going to be a private
cloud that are like how John mentioned
there are certain assets that we could
put on the public cloud it could be
content marketing and certain other
things that could be an industry club by
themselves but anything to do with a pH
I that's a private club
okay excellent
uh this is probably you again for your
regular hours is a medical devices
company with several million lines of
legacy code architected written years
ago with complex architectural
dependencies 20 gigs of binaries nightly
builds take 10 minutes tester extensive
how can microservices and
containerization helping reducing our
bill deploy and testing cycles
what do you have any thoughts are not
anything else they might be able to do
and i'm going to open that up to all
three of you
you see we have a large healthcare
provider here with large large legacy
infrastructure software written directly
for infrastructure does microservices
and containerization offer any help to
them some jeans that might be something
for you sure i think i think so that's
an excellent question and you have some
large companies very large code bases
start the the dog the microservice
journey right so but there are many
factors involved there right it's not a
straightforward you can just float
called detangle it and say I have
microservices sometimes that data
becomes a
you know the normalizing the data to
separate out the persistence components
which microservice you harvest becomes a
more complex effort if you go to IBM
developerworks there's there's an
excellent series of papers by kyle brown
KY le BR all of you and kyle brown one
of my colleagues those that will be a
great place to start to get an idea of
what the effort required is but i would
say that it you it is not you know it's
something but it's a journey which can
be in embarked upon and you take it one
step at a time in and you should look at
by saying how what are the services
which are not deeply embedded in this
monolithic for which I want to expose
first because they can help me in a way
to our speed of changes in somebody
maybe there are certain areas of
application will change very rapidly
there are certain areas which I lose
bother applications very very calm I
already very very often those with the
ones where you would start again it's a
this could be a day-long seminar by
ourselves I think starting with all
those papers were published would be a
great place
also you know feel free to reach out to
me uh you know my contact information is
available on this deck and we're more
than happy to talk to you about how we
would embark on that Johnny video great
i know we're over time I'm gonna take
one last question and i'll put his five
minutes over every she's not stable
I just wanted to add something called a
sad person wants sorry
yeah i think one of the key things in
your question I think which makes it
challenging is your dedicated hardware
if there is a way for you therefore or
ask this question to say is it a way to
use the same piece of code and it seems
like this is a single assembly line
because you have dedicated hardware if
you have the opportunity to have
multiple assembly likes and your pattern
of microservices a pattern just because
you have the ability to launch multiple
been simultaneously and each of those
lines are delivering service
pieces of functionality the codebase
refactoring is a different approach and
also the question I would have is if
it's 10 hours is what it takes then do
you really want to break it and what's
driving you to break it was the reason
you want to break either trying to sell
the 10r problem re-signing trying to
solve the hardware problem which problem
are you trying to stall will determine
whether you to flower or microservices
containers or which piece of lettuce
absolutely so guys one more question I I
think we're gonna have to call it a wrap
i apologize and this is really it's
almost like a multi-speed IT question
how do you deliver a poor call center
solutions in daily releases or even two
weeks every two weeks releases when
back-end systems that give you services
are our monthly or quarterly release and
you know that's kind of at the core of a
lot of the I don't know you son jeev but
you Rosalind in the IBM z systems
enterprise group i mean is what they're
dealing with everyday and John it's kind
of a problem you guys have interacted
hurts right trying to speed up while
still having that back in but who would
like to take a shot at answering that
one I can jump in real quick this is
john just on the surface i agree with
you this you know we were faced with
those questions and instead of trying to
answer them you know or solve for them
and in pieces that's when we realized we
that we have too many of them and that
our problem was the backend systems
first and foremost and then once we got
ourselves free of those then you know
those other front end systems if you
will or those other processes you fell
in line and since we're aligning
business processes to the to be in a
DevOps IT processes in other words we're
not trying to keep our business
processes the same either we're just
transforming it all at the same time
into to avoid that problem and that
wasn't the problem we're trying to avoid
but we did see
that one on the horizon when we're
looking at our options and and you know
saw it as a benefit of doing a full
transformation
ok I excellent china and that does
explain a lot of why hurts make the
decisions they made in your cloud
transformation now we've seen some gee I
don't know if you want to comment but i
I've seen case studies from the IBM
enterprise system where they are they
are marrying these systems of record two
systems of engagement that are working
on different cycles and I often describe
it as more of a Swiss watch where you
have gears that a big ears are small
some gears are spending very fast and
some gears are spending much slower but
they all combined to really give you a
fine timepiece right that keeps perfect
time and I now that's a that's a
delicate operation right which is why
swatches are better than or they say
they're better than watches rate in
other countries but sanjeev what about
what about you what do you say
absolutely nothing you you nailed it
down that a lot of a lot of customers
notice complex it is not easy just like
making a Swiss watch is not easy you
know if you want that never locked
position right if you can go you know
the you know that with a big bang
approach like like hosted nothing like
it call it resolved issues but you have
if you have to go in a more organized a
manner where you're taking piece by
piece and trying to synchronize all
these efforts we have found several
areas that you need to focus on you know
and i'll just list of the top 50 don't
have time to go into each one other the
first one is planning and architecture
right knowing where those dependencies
are doing what does release cycles are
for each one of those delivery pipelines
and understanding the dependencies is
very critical so you're not you know
heading up to wait right you what your
fast-moving application done but then
your wait three months for other
applications to catch up
api is introducing an API economy to
reduce those dependencies and losing
api's like an architectural contract
between different applications and how
they communicate isn't is the second you
know priority area code is and why we
try
one of the Convention which happens a
lot is not just the speed but but
availability of environments right
because you need you need all the pieces
to be there in order test or test
virtualization of virtualizing services
which are not available and organizing
my mess which are not available because
very critical and deploying to the minus
automating the deployments across the
board whether you're doing snow moving
or fast moving again I I have to use the
word speed but you're getting the idea
whether you're a slower cycle or faster
cycle being able to deploy on demand to
the right kind of environments is
critical and lastly release management
right because you can again
hurry up and get stock if your releases
are not managed properly and you don't
know where everybody is in this grand
plan right it's almost like they're
traffic control are you at an airport
have to know very very every airplane is
which one is running out of fuel which
one is it needs to come into the flight
path and they do you can see the
clockwork going on there with planes
taking off and landing you know
simultaneously in multiple run bits so a
lot of planning a lot of understanding
of the architecture
ap is deployment automation and watch
lighting environments are five factors
which need to be looked at in order to
achieve that multi-speed IT which which
you very elegantly described earlier so
that I Gentile handed backhand only i
could go on for this one other hour but
you know we're already way over guys
listen there are one or two maybe three
other questions
I've got to have been riding if I didn't
get your question live here I apologize
i'll submit them to our panel and try to
get you written answers Jules it's on
the back end if we can take care of that
are you know why I wish we could do
these for an afternoon rather than an
hour but we have to be mindful of
people's time
Raju John thank you so much for sharing
your stories with us you know what we've
got to do this again sooner than later
if it's ok we'll we'll talk offline
about it and we'll do a part 2 here and
try to make it even a little more
interactive to allow some of the people
because they have some great questions
Sanjeev is always a pleasure having you
aren't you bring
real life you know so many different
experiences to this discussion it is
truly a a great great thing to have all
three of you guys rocked in on the panel
today for those of us who stayed here
till the end thank you so much for
staying late i apologize i I am I value
your time and I don't need to go over
but i thought it was worth it today guys
thank you so much
this is alan symbol for DevOps dot com
thanks for joining us on this webinar
and we hope to see you soon in another
device dot-com webinar
bye bye everyone
[Music]

Share this subtitle


Description