How to Plan and Budget for 2013 with Cloud in Mind


Uploaded by BlueLockLLC on 04.12.2012

Transcript:
Diana Nolting: Welcome, everyone to "30 Minutes to Success, How to Plan and Budget for 2013
with Cloud in Mind". Thank you for joining us today for this Bluelock webinar. The webinar
will last no more than 30 minutes followed, by a question and answer session. Please stay
tuned to the end when we'll announce one lucky winner to receive an iPad Mini. We'll choose
the winners from the remaining attendees at the end of the webinar.
My name's Diana Nolting, I'm your moderator for this session. Your speaker today is Bluelock
CTO Pat O'Day. Pat's going to discuss strategies, tips and tricks to budget for cloud in 2013.
Please feel free to submit your questions throughout the presentation using the chat
window, and we'll be sure to answer those at the end. Now, I'm going to hand it off
to Pat to get us started.
Pat O'Day: Thanks, Diana. Hello, everybody. I'm broadcasting live from Las Vegas. I'm
attending the Gartner AADI Conference, and I also had time to visit the Amazon Conference.
To put it mildly, the hype around cloud has rapidly transformed into reality because everyone
here, 8000 people between the two events, if not more, are all trying to figure out
how to get this done in 2013, not trying to figure out whether or not they're going to
do it.
That re-emphasized the point of trying to help people figure out how to move forward,
or at least help you round out some of the ideas if you're already doing some planning
going into next year.
Budgeting is, for those of you who have done it, particularly in an IT context, planning,
capacity planning, all of those kinds of things, the level of detail you're going into is really
complex. Getting any of that done in 30 minutes would be a major accomplishment. This is going
to be a fairly high level approach.
I would offer that if anyone wants to go deeper on some of the topics or some of the numbers
that we show and tease out some specific nuances around a particular area of cost, we've certainly
seen those numbers, are very familiar with those things, even what some of the "gotcha's"
might be a little more color around those, please reach out to us after the session.
If there's interest in doing another 30 minute session that takes a deep dive into any of
those aspects, that feedback would be really helpful for us because that's something we're
considering doing.
At the end, we're hoping that you're going to be able to create a solid plan to move
your cloud mission forward, and what you'll see is we've broken the concept of mission
into a small, medium and large, but also the idea of public cloud versus building your
own private cloud versus even using software as a service to some degree.
How to figure out which category you fit into, and then how to take steps forward based on
that general categorization to try and flavor your budget with some actual detail to move
forward. At least know what some of the higher level topics are that might be worth clicking
down into doing some additional research on to make you successful next year.
We tried to throw some numbers out there. We've been doing this for awhile, so we took
some of the costs out of our own environment, some of the costs from some of the companies
we worked with that have gone and built their own environments and some of the areas of
cost within each one of those larger buckets, [some] things to consider.
Then, we also tried to list, at the end, some "gotcha's" and some other things to consider
because if this is new to you, there are a lot of potential pitfalls that might catch
you, and that's never good when you're trying to explain to the CFO why you missed a fairly
large expense or when you're trying to explain to the business that it took you three months
or however longer, to get something done than you originally planned. We're going to try
to help you avoid as much of that as possible.
What happened in 2012? Get this going even further. Like I said, the interest out here
in Vegas right now is tremendous, but let's try to turn it into numbers. Clouded option
increased to 48%, so across public cloud, private cloud software, the service based
on this data, it looks like almost 50% of companies are using it in some way, shape
or form and that's a lot.
I don't know what the trajectory looks like specifically, but it's to envision that by
the first quarter, second quarter of next year that more than half of companies will
be using cloud in some form. Eighty three percent of IT decision makers believe that
it's been effective.
Short story is those implementations occurred and they went well. They solved a problem,
or a specific problem the company had had in a way that they hadn't been able to solve
that in the past. They're going to go further. Their adoption is going to increase.
We've seen that. We've seen companies come into our environment just more in a tire kicking,
proof of concept mode and then really make big investments into it later because they
get a rhythm and understand how it works.
A 27% rise in first time users, so you look at the number of people that were reading
Gartner research reports, reading Internet articles and not ready to do anything, this
is a big spike up in the number of people that have taken a step forward and have given
it a try in some fashion.
In terms of picking your approach, this may be too distilled, but we thought it would
be easy not knowing who would be attending the call, and I imagine a lot of you are either
from small companies or potentially even from larger Fortune 500, some of you might be business
users, some of you might be in IT.
We wanted to do a general purpose thing that you could start with to land somewhere. Then
we're going to tell some stories about some companies and hope that you identify with
and where we've plotted out where they landed here.
How do you get started? This is the real key. There are three ways we're seeing people approach
this and they have a huge impact on what the spending would be. The first concept is this
walk before you run.
If you're familiar with virtualization and you're using it in your environment, you may
have already virtualized 25%, some people have virtualized even 50% or 75% of their
environment, the idea that you're going to take that next step and convert your virtual
environment into a private cloud by adding self service, by adding API, object storage,
a couple more capabilities, but you're all in this progressive journey moving forward.
A very thoughtful approach to it, very iterative, five percent of the time versus full change.
The other thing we're seeing is that people who aren't ready to do that, or maybe in 2013
they may end up with the funds to convert their environment into a private cloud, but
the business units want to start acting with agility, even right now, but certainly when
their funding gets approved in January, if you're on a fiscal calendar.
They're going to want to start working on cloud or acting in a cloud like way immediately.
In these situations, it's not so much about what you're doing with your infrastructure
from an IT standpoint, it's about specific projects that these business units might have
and how do you enable them.
We have seen a lot of this where a company will pick a marketing initiative, maybe a
new application, and instead of spinning it up somewhere in the data center somewhere,
they'll actually use that as a way to learn about cloud and deploy that.
Some companies are literally rethinking everything. They're trying to transform everything they're
doing to more of a cloud first mentality. These are folks that have typically experienced
cloud, have been using it for two or three years and have realized the benefits are there,
they're very real.
The benefit is much greater the faster they can convert to this new normal. That's also
if you don't understand what the benefits are and what the risks are very well. This
could be, potentially, really detrimental if you took a little too much risk.
The other piece, we talked a little bit about this before is who is being served by this
change most effectively in terms of budgeting? If you're an infrastructure owner in the IT
organization, you're used to buying hardware, software, maybe even power and cooling, depending
on what your role is there. You're really traditionally measured based on efficiency,
and you're really trying to be all things to all people. A lot of conflicting priorities.
Your world, typically, from what we've seen, revolves around what we call general purpose
infrastructure. You want a general purpose cloud environment that can serve any number
of business unit needs or application and data needs. Different levels of security,
different levels of performance, but all coming from one environment.
Application owners aren't always in the business unit, they're sometimes in IT but their real
focus is around one specific, or maybe a handful of specific applications, and they're looking
at all these agility benefits, paying for what you use, being able to get resources
the minute you click the mouse.
Their challenge is they're not always a priority to IT. They're a priority, but they're not
always necessarily the number one priority relative to how quickly they need to get something
done.
It's like looking at a Starbucks line as you're driving by in your car, walking down the sidewalk
and you've got two different coffee shops and one's got 20 people waiting in line and
one has nobody, or you go to the kiosk and put your coins in the machine and get coffee
out that way.
The ability for a business unit to make that decision as to whether they want this mission
critical good stuff and they'll wait in line to get that with everybody else, or they just
need to get something done immediately and they're willing to take a better risk, so
they get that done.
Let's get into the dollars and details now that we understand a little bit of the context.
This may be a little backwards, but we're going to give you the numbers first and then
we're going to back into how we got there. If anyone has any questions or wants to go
back to that at the end, let us know.
This matrix shows the small, medium, large at high level buckets of spending, and we
will click down to show you what the details are in each one of those buckets, and in the
left to right is public cloud, private cloud and software as a service.
Once we get into the slides and you pull all this together and you can see where you're
going to land in terms of your spending, you can identify which bucket's here. In hindsight,
we should have put this at the end, but that's my fault.
For a small, let's go to the upper left corner, so the small public cloud environment, you're
looking around $3000 a month. This will handle a handful of applications or one smaller Internet
facing application, all the way to, some people are spending $10,000 to $100,000 a month,
but you're looking at that in contrast to what might have been a $5 million or $6 million
capital spend.
It does sound like a lot of money, but when you're looking at what it would have cost
to build a private cloud part of it, it all works itself out. Everybody's buying the same
components at the end of the day, RAM, CPU storage and network bandwidth, so there's
not a lot of mystery in that.
Private cloud, the key nuance here is how much capacity you need, and we'll show that
on one of the next slides, is how much capacity comes out of each one of these numbers. You
can invest anywhere from something as simple as $10,000 or $20,000, buy a pair of servers,
a little bit of shared storage, install VMWare on there and install some cloud capability
like VCloud Director, and you've essentially got 100 VMs or so, 20 or 30 VMs or so, worth
of self service you could offer to a business.
For a small organization, or remote office, that might be pretty effective in giving them
some independence, but keeping the IT organization in a role of governance.
All the way on the right side, $500,000 number is probably not even realistic. It's probably
$2 million or $3 million. I know that just the storage environment that we buy when we
light up the new cluster is more than $500,000, and that's without any compute, without any
network, no cabinets, no power and cooling.
The other key is if you look at the contrast here with being a private cloud line and a
public cloud line, you'll notice that there's two categories of spend on private cloud,
and there's only one on public cloud.
That's this operational dollar, so the monthly recurring cost for the labor, power, any kind
of software licensing that you would have, all that stuff that repeats, you're paying
those costs independently in a private cloud scenario, but they're actually folded together
in the public cloud.
One mistake we see is when people go to budget for spending is that they will, on the private
cloud side, they will only budget for the hardware and they'll miss some of the operational
costs.
Or, if they're trying to figure out whether they should invest more in public cloud versus
private cloud, they will only compare the cost of the hardware. They won't actually
include the cost to operate all of it, which is essentially free in the public cloud side.
If you just look at it as a hardware cost, it's invisible in there. Just know that we
have no magic wands here. There are real people doing real work just like there are in your
world to keep those assets running.
That same concept applies if you go down to the bottom line with 'software as a service'
where you've got more of a per user cost in most cases and that also includes, not only
the hardware cost, but it also includes the people and the process side of it, patching
software, the security.
You think about, 'How do these software as a service companies protect themselves when
they come under massive denial of service attacks?' They spend money on really good
technology to stop that and that gets folded into that per user cost. It's important to
tease those things out, especially if you're talking about doing this at scale and really
trying to do any kind of decision about where you put a workload based specifically on cost.
What will these spends get you? I'll jump through a couple of examples here. Let's go
down to private cloud line in the middle. As I mentioned, if you look at it from a VM
count, the easiest way to get here is to take how many virtual machines you think you might
have because you're going to be enabling self service as part of this process, how many
VMs do you think people might spin up.
Figure a factor of how many gigs of RAM, how many virtual CPUs, how much storage the average
VM already in your environment uses today, and then multiply that all together, and you'll
get a rough feel for how much storage, how many CPUs you need.
You can then go approach your typical hardware vendor, whoever you currently use, CDW, Dell.
They're good at putting a packaged offering together for you on the small side.
On the large side, when you get into 800 terabytes, thousands and thousands of VMs, services,
you've got backup and all these other things that need to integrate in there, there's a
lot more complexity involved in that.
You start to ask little questions like, "Are we going to back the entire private cloud
up, or are we going to allow the users to come in and pick what gets backed up on their
own?"
That will change how much backup capacity you need. The biggest thing in private cloud
that we see people run into is this concept of 'sprawl'. If you're sizing this based on
what you think will happen, I would encourage you to at least go to 150% of that, if not
200% of that because once you open the floodgates around self service, and there really aren't
any controls in terms of cost.
That developer can come in and spin up 5 machines, 10 machines, 15 machines and they don't see
impact of what that costs the business. They're going to do that to try and get their job
done which is certainly a healthy capability to add, but if you don't have any way to manage
or constrain that from a cloud standpoint, just know that sprawl is going to occur.
If you look at the public cloud line, we see people going in with tests and dev workloads,
kicking your tires, that's what those numbers that we threw out were. You can get cloud
resources now for a couple of dollars a month up to a couple hundred dollars a month, but
it's little tiny environments.
There are people... I saw a great presentation from NASA yesterday, running large mission
critical workloads, big high availability, guaranteed performance, high security requirements
in the public cloud. Their challenge is, from a cost standpoint, they put in a lot of investment
into the cloud resources themselves.
They also had to put a lot of people in process and even additional technology on top of their
cloud span to achieve all these capabilities because in a lot of cases, the cloud itself
doesn't give you as much security or as mission critical capability that you need, or it gives
you all the tools, but you've got to add additional technology to put all that stuff together.
When they were talking about the cost on the prior page, those are the things we had in
mind.
If we go quickly to the SaaS line, you just need a few users, a Dropbox sort of scenario.
They just need to collaborate. It's not that big of a deal, but as you start going into
larger investments, and we'll talk about one of those, the amount of integration costs
that you're going to pay.
It may be $25.00 a user to get access to the SaaS application, or CRM or something, but
to make that CRM environment behave the way your business behaves may require a several
hundred thousand dollar, if not a million dollar investment from professional services
because they've got to do all the integration of your business process.
Some budgeting in action. These are three stories. This is a SaaS story. This company,
a large pharmaceutical company that we work with, they're doing investments in SaaS, in
public cloud and private cloud. They really wanted to build a large private cloud environment.
I think they launched that this year because they want to transform how all the VMs in
their internal environment are being used and modernize their existing infrastructure.
Then they also wanted to modernize some of their applications, and when they looked at
their existing CRM solution, it was supporting about 7000 sales people, they realized that
the easiest way to do that would probably go to a SaaS offering.
Instead of trying to transform or modernize the infrastructure and then modernize the
application inside of their data center, they actually moved the entire application outside
of the building and put it with SalesForce.
The other thing that they were observing was that even though they had this private cloud
infrastructure, they couldn't turn resources off to save money. Even if a business unit
went in and turned off 100 virtual machines, or five or ten virtual machines, they couldn't
give the business unit relief from a cost standpoint.
Even though all those machines had been turned off, other than a little bit of power that
wasn't being consumed, IT still had all the same resources in their data center. They're
using public cloud at hybrid strategy because they know that with the public cloud resources
they can turn them on and off.
They can get 100 machines for a month or for several months and then shut that off when
the project is over and immediately realize the savings. If they have an application that's
going to be around for three or five years, they actually spin that application up in
their private cloud environment because they can depreciate it just like they would physical
hardware.
This was our matrix. If we look at this particular pharmaceutical company, you can see that from
a small, medium and large they did a small public cloud environment to handle these 'bursty'
workloads that they can turn off and turn on really quickly.
They also made a really significant spend on the private cloud because they're modernizing
their entire VMWare infrastructure and they outsourced a huge CRM environment into SalesForce
to get agility to add to that team.
This is another company that did one thing to buy themselves time, and then did something
else as they moved along. As a large retail grocery chain their vision was very similar
to the pharmaceutical company that we described before in that they wanted a hybrid public/private
cloud environment.
They knew it was going to take time over 2012 to get their private cloud built up and running
and they wanted to enable the business units back in January to get to agility. They started
off with a medium sized public cloud investment, AR environment and then over the period of
the year they built a really large private cloud environment.
when that environment got up and running, got through quality testing, they actually
moved their medium and public investment down to a small public investment because they
were able to move all the applications they had spun-up through the year, back into their
private environment, but they kept a small public environment up to handle the diversity
workloads that they couldn't predict very well.
The third example that we have is Siemens, their eMeter group. They took more of a stratified
approach to this. They wanted to get different levels of resources in the cloud, and they
started off with production put in the cloud environment, and they ended up moving testing,
and they ended up moving QA and all these other components. They basically cloudafied
everything.
They run it in more of a hybrid approach where most of their assets are in the public cloud
because the users that use their applications are around the world, so it's a lot of Internet
facing stuff, and it is production, but they also have resources in their VMWare environment
on site when they just need quick turnaround and churn in getting some things done internally.
Then they move assets back and forth as needed.
There's some exciting animation there on that slide. In terms of planning your approach,
I know we're running up on time here, and we're almost done. I want to make sure we
have enough time for questions. The other key is, let's say you have an idea now what
market you fit in, which perspective you're looking at the world from.
Are you in infrastructure? Are you an application owner? You're a medium- sized company and
you think you're going to do mainly private cloud but you want some 'burstable' capability,
and you think for collaboration you might turn on Dropbox or Box.net and enable business
users that way, or maybe Evernote.
Wherever you land it, as you begin your journey, once you've got your funding, the next step
is figuring out what workload makes sense. You heard some examples as I went through
the different stories, but in every one the case is different.
If you're in the, "I'm transforming everything I do," it might be worth looking at low risk
stuff, low hanging fruit. If you've got old Lotus Notes infrastructure that's already
virtualized and you're just trying to free up some capacity on your high performance
SAND that you've invested in for your private cloud so that mission critical applications
have more resources, it might be worth taking some of the old stuff and actually moving
it into the public cloud to free up resources for all the new growth that you're going to
go through in next year.
Or it could be the 'Pick a Project' approach so the workload is a specific business unit,
they come in hot and heavy. Maybe they didn't call you to plan as they should or as you
would have hoped 90 days, or 120 days out, but they need it now. You can still take their
growth out into the public cloud [via] your private cloud deployed, and bring them back
in-house when you're ready.
Get that thinking there. In terms of the research that you need to do, there are a lot of cloud
providers out there. There are a lot of private cloud technologies where you're looking at
Citrix, or Microsoft, maybe Eucalyptus, VMWare, which is what we use, figuring out what technology
might make the most sense for your organization, for the skill sets that you have, then some
form of evaluation and pilot.
Fortunately, most of the public cloud providers offer an easy way for you to test drive their
infrastructure. Private clouds are a little harder to test drive. One of the neat things
we see people do is if they think they're going to deploy VMWare based private cloud,
they'll do a test drive in our environment because it looks exactly the same as what
you would get if you deployed it on premise.
They learn whether or not a VMWare based private cloud would work for them by using a VMWare
based public cloud and it helps them understand whether that makes sense to go forward.
Other cost considerations. As I mentioned before, you may need help if this is the first
time you've done this, getting through that journey, getting through that transformation,
so there are professional and advisory services available, whether you need that on the business
side to understand which workloads might be the most optimized, or might become the best
workloads to move into a cloud-like environment.
Of, if you just need a one-time deal where somebody comes in, drops the equipment off,
gets it all set up, wired together, configured, it helps you get through that first implementation.
Training is available today whether learning the cloud infrastructure technology itself,
if you want training from VMWare or any of the other infrastructure providers on how
their cloud technology works. A lot of cloud providers now are offering training in terms
of how to use their specific cloud environment.
Stop gap resources are important. Taking some sort of percentage of what do you think the
spend would be, and just imagine that you may run out of capacity or you may need more
than you need and being upfront with your CFO in terms of planning and say, "This is
my best guess. I think we're going to have 500 VMs on this environment," but if it turns
out it's 700 or 800, at least you've already gone to the CFO and said, "It worked a lot
better than we thought. People really liked it and here's what I think the next spend
would be."
Some thoughtfulness in terms of your planning helps there too because what we've seen happen
is a lot of people will put in a private cloud infrastructure that's not expandable. The
idea of buying a storage controller that you may only need two ports on it to hook up two
servers or four ports to hook up four servers, maybe go ahead and spend the money on the
eight port controller or the 16 port controller so that you don't have to buy a whole new
controller and a whole new setup and build it all over again just because you had one
more VM than what your existing product [inaudible 29:42] would support.
Something else we're seeing, we're seeing people use public cloud to dodge that bullet.
Let's say you do fill up your private cloud environment sooner than you expected, and
it's going to take 30, 60, 90 days to get more capacity set up and running, you can
drain the swamp a little bit by moving some of those resources out to the public cloud,
buy you time to get your private cloud expanded, and then bring the resources back.
We talked about sprawl a little bit, but these are all things to help you save time and money
and manage expectations within the business.
We talked about adding a trial amount of money in your budget so there are free trials, but
if you try to put a real application out there and exercise the cloud, it's going to cost
money.
The idea of a phased approach and spending and say in Q1, you put in 25% of the spend
due to if you decide you're going to do private cloud.
Then you'd employ the majority of your spend if your trial went well, or you increase your
public cloud spend a little bit more than the trial. Rethinking everything, a drastic
amount of planning in that regard.