Tip:
Highlight text to annotate it
X
x
Hi, my name is Soeren Juul Schroeder from Schneider Electric.
Let's dig into some of the themes of DCIM.
These are industry themes, not necessarily related to any products or modules,
more related to the kind of vibes you will be getting from customers what you'll see at conferences etc.
So think of these themes as very generic,
addressing an area of the customers environment and his key concerns.
So let's start looking at those.
The first one will be around monitoring and control and the obvious questions you could ask yourself
when you're addressing this
what you typically would be asking the customer
and what the customer would be asking himself could be -
Who gets notified when a UPS goes on battery?
And this is not really to get a specific answer out it
it is to get a discussion out of it around why does anybody need to be notified when a UPS goes on battery
and what is actually the impact on the business.
Very few companies out they have a clear idea about what is going on when something happens in the data center.
What will the potential impact be?
To give you an example of monitoring and control -
This is monitoring - this is the dashboard of a very old car.
The monitoring is really limited to how far have we been going and how fast have we been going.
It is very basic.
Additional monitoring probably happens - but it happens between the ears of the guy running the car.
So typically - something doesn't smell right, or something doesn't sound right,
that's also monitoring.
This is really how most of the customers do monitoring in their data center today.
We're suggesting to take monitoring to the next level,
start talking about in terms like this -
this is also a dashboard from a car, but a fairly new car.
If you look at the far right you can see
approximately how many liters of gasoline or diesel you have
in your tank right now, to support your engine.
That is not really the exact amount
that's beeing displayed.
What you need to care about instead is really what can that do for you.
Behind this dashboard is actually a computer. That computer translate
how you've been operating this car so far
and how far can you keep operating.
So you have a range approximation
instead off the typical x liters or gallons left in the tank.
It is really taking a monitoring from the device based monitoring into a context-aware monitoring
and that's what we need to do in data centers as well.
Next topic is about virtualization.
A question you could ask yourself is -
Based on which criteria do you decide virtual server moves?
And again this is really interesting to get the discussion on
why do you move virtual servers,
this is happening very often in many organizations,
depending on size obviously.
But what we typically see is the operator of the virtualization tool belongs to an IT group.
He has little to no knowledge about the actual well-being of the
physical infrastructure in that datacenter that supports his virtual operation.
He typically makes his mind up based on
the processor the memory and storage.
Why is that?
Well it's typically the parameters he can see via these virtualized consoles,
such as VMware, Microsoft etc.
What we all know, the ones who live in the data center world,
the physical infrastructure and all the events that can
happen around the physical infrastructure,
has a huge impact on the wellbeing of the pbysical infrastructure as well.
What we need to consider is how do we tie health status
of the physical infrastructure with the health status of the virtualized environment.
Those two things need to be working hand-in-hand.
An example: an event happens.
That event has a business impact on the physical host servers for the virtualization
and the guy that operates that really needs to be made aware.
If he is trying to move a virtual server
to a physical server sitting in a rack that is not in good shape, he needs to know that,
and by integrating virtualisation and DCIm you get that capability.
Next up is visibility and transparency.
The question you'll ask yourself is -
How do you know the impact of a data center event?
How do you really know what's going to happen? Similar to what we just talked about,
both regarding virtualisation as well as monitoring and control,
I like showing this picture to people, because it describes the situation perfectly
that many organizations are in, when it comes to managing the data center.
It is taken from the story of the blindfolded men and the elephant.
You can clearly see
none of the people, all touching the same elephant
are actually capable of telling what is in front of them,
because they are all blindfolded and they are unable to use all of their senses
and this is one of the critical parameters in the data center today.
The facility team is looking at a certain set of parameters,
the IT team is looking at another set - typically happening in silo'ed systems.
You really need something that can bridge the gap between the two
and that's where DCIM comes into play.
Another theme and this is probably one we're going to be talking about at conferences in the very near future -
resource balancing.
The question you want to ask yourself is -
When will you run out of capacities and how would you know?
First of all it is a very hard question to answer.
You probably need to call a couple of guys.
You probably need to look into a couple of systems to get some kind of answer to this question.
When I ask my customers this question I typically get a lot of pushback -
What do you mean capacity, which capacity, what are you specifically talking about - is it power, cooling?
But really it's every single capacity
that can prevent you from delivering whatever availability, uptime or whatever
that your SLA is describing.
And why is it difficult?
Remember there's a lot of different components in the data center, there's a lot of different people
or groups that owns various components
and again the management of those components can happen in various systems.
However, no matter what change you're doing in your data center,
it will have an impact on something else. It can have a direct impact for instance if
I try to deploy a virtualization blade chassis
what will happen there is it will have an impact on my power draw
in that rack so hence it has a power draw challenge
to the UPS and needs to be able to deliver that.
It also has an impact on the cooling, because it's heavy on power draw,
it will be delivering heat back to the data center,
so there's a direct impact on the cooling as well.
If the cooling is on the powertrain of the UPS it will have an
indirect impact on the UPS again.
So you need to get in control of all this resource balancing
because having enough power and cooling is critical for you.
Having too much power or cooling can be even more critical
if you're actually trying to run an optimized organization
where the actual energy spend is a key concern for you
and without having a tool that can bridge the gap and help your balanced all those resources
you would really be running blindfolded again.
Let's move on to the next theme which is automated capacity planning.
In your organization, who decides where the next server goes?
The answer we're looking for is really
How do they decide where to put that next server?
It seems like a very simple operation - it's just a server
needs to go into a rack, needs a little power, network etc. and that's it!
But there's a lot of impact by doing that.
There is the obvious impact - maybe there's not enough space in the rack you had intended.
aybe there's no network ports available in that rack.
Maybe you even get into a situation where you trip a breaker in the PDU
by powering on that server - not very good,
but you will spot it immediately.
However there's also a lot off hidden implications in this scenario.
You have designed your data center for some backup time
on battery that will allow you to power up the generator
in order to deliver power back into the data center.
Maybe you designed that for five minutes, because you need three minutes to get the generator off the ground,
but maybe the way you using your data center today you're down to only two minutes of backup time.
That's a huge problem as you won't have time for you to key off that generator and you won't find out till the power drops.
Another area that might have hidden implications is around cooling.
Again you have designed that in terms of redundancy and capacity.
How can you verify ten years later that you're still capable of delivering enough cooling into it.
The overheating doesn't start immediately when you deploy a server.
It can start weeks or months after - you never know when that hot spot arises.
When it arises it's too late and you are in a critical firefighting mode again.
So, what can you do to prevent yourself
from getting into a situation where hidden implications come back and hurt you.
Another theme we can talk about is workflow management.
A typical question you can ask yourself here -
What unauthorized changes are being made to the data center that could impact availability?
Why is that critical?
I had a customer once who described his data center like the wild, wild west.
He never knew what was going on, who was coming in, who was removing stuff, was moving things around for various reasons.
And the problem was every time something broke down it was his problem,
even though he didn't have a clue about who did what and when.
So getting in control of that is very critical and through a structured process where
somebody plans it with good data at hand,
somebody authorizes it and somebody executes, but all getting captured
in a single pane of glass. That's again another theme of DCIM
and it needs to be covered.
Efficiency can be many things. One question you could ask yourself here is -
How efficient is your data center?
Thereby also, how efficient is your business?
Because there's very often a one-to-one link between the two.
When we talk about efficiency it is a funny size because,
what is efficiency really? We can talk about operational efficiency
making people more efficient at what they do. Taking some of the manual tasks out of the equation
and trying to automate that.
But it could also be plain, simple energy efficiency,
optimizing the energy spent around your data center.
Unless you really capturing these things and really have a tool
that works around it, you're in a very difficult situation.
Next thing we will talk about will be reporting.
A question you could ask yourself there is - How do you communicate your data center critical metrics?
This was a very big surprise to me
that capturing data in the data center can be troublesome
and also distributing it up to the right receivers can be even more time consuming.
A lot of time is spent in this category on a daily and monthly basis in many organizations.
It's again another candidate for automation, where you really
dig in, define the critical KPIs for a team, or multiple teams
and you work with that KPI and you make it very visible
to the receivers.
You know what the status of that KPI is. Just remember, KPIs like anything else,
they change over time. Static systems that just delivers a predefined set of KPIs
they are not good,
that system needs to be very easily
integrated ino the workflow of it. So you can easily change your KPIs to whatever
is the KPI of the month or the year.
So again, another theme that is very interesting to talk to customers about.
Next theme is mobility and the question you could ask yourself or your customer is -
How do you stay connected when on the move? Again we all know this
nothing is static and we keep moving around. Things come in, things go out
and we as human beings move around.
Having access to your critical data, your KPIs, your critical parameters around your data center,
the well-being of it righth now, the criticality of any issues and challenges coming up,
it's key to have that easily accessible.
This is where you get into discussions about having the data available
wherever you are, whenever you are, whatever device you're using
whether that's a laptop, a mobile device, a smartphone or tablet.
With that the critical data needs to be available to you.
Last theme that we will talk about is implementation.
An obvious question you can ask the customer here is -
Do you have surplus time and resources? Many customers are already pressured on headcounts.
just imposing 3 months of activity on them for actually getting control of the data center
might be enough for them to say
I know it is the right thing to do but I can't find the resources right now.
Which is why we will go into a discussion about
the value of a certified engineer, who has done this many times before, with many similar customers
and maybe even similar verticals or countries.
It has big value to the end customer, knowing that he's in good hands
and he'll be taken well care of.
So that was just a couple of themes
that we can go through
as part of our discussions with customers, without really getting into the product discussions.
Themes that are so generic
that you put your own stories around it, find your own anecdotes
and start to talk around these themes with customers.
They will love being directed to something that sounds really good
and comes from a knowledgeable person that has done this before.
Use these themes before you even consider going into product mode.
I hope this was beneficial for you - Thank you!