Microsoft strategy going forward-- how will this affect certs?

Some of you have probably heard that Microsoft is now on a path to becoming a "devices and services" company.

Do you think this will see them start bailing out of local server platforms, and along with that certifications?

If they do get out of servers, I'd have to think open source platform certifications (e.g. RedHat, Cloudera) would start taking on more prominence.

What do y'all think?

Find more posts tagged with

Comments

iBrokeIT

Their direction is a bit uncertain at the moment since Balmer is on his way out and they haven't selected a new CEO. Personally I think they should stick to what they do best which is software instead of trying to compete with Apple which has turned out flop after flop: Windows Phone, Zune and Surface..

ajs1976

Microsoft is trying to become 'the' cloud provider and they are throwing a lot of money at it. Even if that does happen, I don't see the OS and applications going away.

Once a new CEO is chosen the plan may be adjusted some.

Asif Dasl

Mark Minasi has hinted that Active Directory is heading towards the cloud, so I don't think you will see the evaporation of certifications. Amazon Web Services has certifications and I can see Microsoft going that way for maybe the next OS release in the next 2-4 years time. There is talk about the company splitting between business & consumer products, that may happen also to become "nimbler".

AFAIK there are no stats on server installations other than web servers (of which linux takes 65% of that market) but I can't forsee them bailing out of server OSs, just moving to the cloud more-so. If you were to believe Microsoft's own figures they say 45% of IT spending will be on cloud services by 2020 (click on "Transform the datacenter") then there must be huge ramifications for Server OSs coming down the pipeline.

googol

I have too wondered about this, as I progress with VCP and MCSE: SI for Server 2012. Makes me wonder what can I do to future proof myself, obviously not entirely, but at least give me some sense of stability and skillset for the future.

It is starting to seem like I need to shift away from Microsoft in some cases, but I do not see companies switching anytime soon, like previously mentioned..2020, but that is going to come quicker than some realize.

kj0

Ahh.. 2020, If our opposition leader is elected this saturday to be our Prime Minister, 2020 will be when we are able to receive 25Mbits to our doors.(A slight upgrade from our 24Mbit ADSL2+)

Looks good when so many business' are moving to the cloud. Everyone will be having a downpour from the cloud(s) and we will be getting the occasional spit.

However, M$ does hold a fair share in Domain Controllers and I think it is due to the ease of AD. ConsoleOne isn't too bad, but I do prefer AD. When I rollout Mac environments I still prefer to install a Windows DC to house all their accounts/profiles/Home drives just for ease. I would hate to see M$ take away the server OS and lose this ability.

What would Bill say about this direction M$ has been travelling lately?

RobertKaucher

I think a lot of people seem to see "the Cloud" primarily from a SaaS provided by large firms stand point and that is really quite narrow.

Private and Hybrid cloud is what will change everything. And I don't mean that in a mooshy vague marketing stand point. I really mean that what we as IT pros and developers will be providing to the company is going to change significantly. A company's internal IT will become the cloud provider or they will be seen as a road block to be worked around.

Windows Azure pack

What’s New in 2012 R2: PaaS for the Modern Web

pram

'Private Cloud' is literally a meaningless marketing term. I challenge you to point out any major difference between a Private Cloud and provisioning servers with VMWare or Xen etc. You have no platform elasticity for scaling, you're still using the same big iron as before. Your redundancy still comes down to having the same exact SAN hardware and geographic datacenters.

So what exactly is changing and where does the Cloud part come in.

dave330i

pram wrote: »

So what exactly is changing and where does the Cloud part come in.

Self service portal & life cycle governance.

pram

yep those are real game changers right there

dave330i

pram wrote: »

yep those are real game changers right there

Perhaps you're in the wrong line of business.

Claymoore

How do you provide Services without the Servers to Serve them?

In the end it's all about users accessing the services they need to get their job done. They don't care if it's on a local physical server, a remote virtual server, or a hosted app in a cloud somewhere. We care, but they don't have time to worry about such things. The server OS will get more complex over time to provide the flexibility to meet the business demand.

Look at all the features that have been crammed into the server OS in the last decade:

Terminal Services / RDS
ADAM / Lightweigt Directory Services
AD Federated Services
Hyper-V
Deployment Services
Lots of improvements to original NT/2000 services

The server OS isn't going away. It's changing faster than ever.

RobertKaucher

pram wrote: »

'Private Cloud' is literally a meaningless marketing term. I challenge you to point out any major difference between a Private Cloud and provisioning servers with VMWare or Xen etc. You have no platform elasticity for scaling, you're still using the same big iron as before. Your redundancy still comes down to having the same exact SAN hardware and geographic datacenters.

So what exactly is changing and where does the Cloud part come in.

Please allow me to introduce you to the National Institute of Standards and Technology and their definition for Cloud Computing. So unless you believe that NIST is a marketing organization well, then there is a standard, technical definition of what Cloud Computing is.

NIST Computer Security Publications - NIST Special Publications (SPs)

Here is a summary of their definition:

Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model is composed of five essential characteristics, three service models, and four deployment models.

On-demand self-service: Non-administrators are able to deploy servers, storage, databases, and other resources without going through corporate IT processes that can some times take weeks or months to provision new systems due to onerous processes and procedures that increasingly make IT Operations seems like a road block to other departments. Marketing needs to spin up a new site for their next campaign? They just go a to a policy driven portal and click, click they are ready to go. Developers need to spin up a new test lab to do root cause analysis on a failure that is occurring at 4:00 AM when no one is around to approve the request for 6 new servers? Go tho the portal, select what you need from the available templates and go.

Broad network access: BYOD and other device integration strategies come into play. Users should be able to get their work done regardless of where they are or what devices they have at the time (within reason of course).

Resource pooling: Departments that have spurts of processing utilization can now take chunks out of the available resources as they need it. So if your finance depart only needs a certain amount of computing resources for a week during quarter close, then they can get it on demand and those resources can then be returned automatically to the shared pool. This is a very different model where in the standard model finance usually has its resources allocated statically and continue to cost the company money even when not in use.

Rapid elasticity: This goes along with resource pooling in that if demand on a resource increases it should not fail, but scale automatically to handle the increased demand. For many services it will appear to the consumer that resources are unlimited and can be apportioned in any quantity at any time.

Measured service: Here is where we see a big shift for IT departments. Instead of acting as a black hole into which money is thrown, IT departments can now report out on a service basis what each department is using, how they are using it and what it costs.

Now let me attack your complain that we are still using the same big hardware as before, that we are still limited by the same SAN and geographically confined datacenter. If you are still doing things the same way as you did in the past, but just using virtualization, then you are by definition not doing private cloud. If your developers are not able to go to a self service portal to provision their own systems but instead have to go through a drawn out process to have someone on the ops team build them servers and purchase licensing, then you are not doing private cloud. I'm not saying everything has to be self-service and scripted/automated. But that most of it is that way is the point of it. A distinct shift in what we are providing to the enterprise and how we are providing it and tracking it. What I see as being the biggest changes for my organization is the self-service aspect and resource pooling so that one area can scale up on demand rather than having to purchase all the capacity they will ever need (up to a point) up front. So the idea that QA and Finance (who are resource hungry but only in spurts) can do much more because it actually costs them less since they are not purchasing all their resource in a dedicated model now. I don't need two dedicated servers with 32 GB RAM and a massive amount of storage to process their cube/ETL. I just have a smaller machine running the cube in production and then when closing week comes, a new system is scripted to come online that hooks into the ETL system, someone on the Finance team pushed the button saying the books are closed and boom! The job gets started and when it's all finished the extra system for ETL disappears, I never had to touch it, and the amount of RAM used on the production server running the cube drops off, everyone gets their stuff. My job is helping the server guys on my team automate it all. We are two "admins/DBAs" and two "admins/developers" on my team. We do a lot of automation and working with users to understand their requirements.

pram

dave330i wrote: »

Perhaps you're in the wrong line of business.

uhh

RobertKaucher wrote: »

Please allow me to introduce you to the National Institute of Standards and Technology and their definition for Cloud Computing. So unless you believe that NIST is a marketing organization well, then there is a standard, technical definition of what Cloud Computing is.

NIST Computer Security Publications - NIST Special Publications (SPs)

Here is a summary of their definition:

We're not talking about the Cloud. Cloud and Private Cloud are not synonymous.

Resource pooling: Departments that have spurts of processing utilization can now take chunks out of the available resources as they need it. So if your finance depart only needs a certain amount of computing resources for a week during quarter close, then they can get it on demand and those resources can then be returned automatically to the shared pool. This is a very different model where in the standard model finance usually has its resources allocated statically and continue to cost the company money even when not in use.

Quite literally starting a server in Xen. You're using a shared iSCSI pool. Shared CPU. Delete the container and surprise those resources are available again.

Rapid elasticity: This goes along with resource pooling in that if demand on a resource increases it should not fail, but scale automatically to handle the increased demand. For many services it will appear to the consumer that resources are unlimited and can be apportioned in any quantity at any time.

How can a private cloud be elastic? Will each corporation simply purchase spare hardware? What happens if you exceed the hardware you have sitting in your datacenter? Are companies that are averse to large hardware purchases suddenly be willing to buy spare capacity simply because its now a Private Cloud? Doesn't sound very elastic IMO. If you shut down containers you're still paying for the hardware, electricity, etc etc.

Measured service: Here is where we see a big shift for IT departments. Instead of acting as a black hole into which money is thrown, IT departments can now report out on a service basis what each department is using, how they are using it and what it costs.

Now let me attack your complain that we are still using the same big hardware as before, that we are still limited by the same SAN and geographically confined datacenter. If you are still doing things the same way as you did in the past, but just using virtualization, then you are by definition not doing private cloud. If your developers are not able to go to a self service portal to provision their own systems but instead have to go through a drawn out process to have someone on the ops team build them servers and purchase licensing, then you are not doing private cloud. I'm not saying everything has to be self-service and scripted/automated. But that most of it is that way is the point of it.

No you're still using the same hardware as before. All you said is its going to have a portal to spin up containers. Xen has templates, everything has templates. No one needs to build machines unless you're running everything on bare metal lol.

So the idea that QA and Finance (who are resource hungry but only in spurts) can do much more because it actually costs them less since they are not purchasing all their resource in a dedicated model now. I don't need two dedicated servers with 32 GB RAM and a massive amount of storage to process their cube/ETL. I just have a smaller machine running the cube in production and then when closing week comes, a new system is scripted to come online that hooks into the ETL system, someone on the Finance team pushed the button saying the books are closed and boom! The job gets started and when it's all finished the extra system for ETL disappears, I never had to touch it, and the amount of RAM used on the production server running the cube drops off, everyone gets their stuff.

You just deleted a virtualized container. What happens if the needed capacity exceeds your smaller machine? How elastic is having to order new hardware? It isn't.

I mean I guess if the point is there's now a friendly GUI for starting templated virtual containers then I totally concede that point. That isn't a profound change though.

LarryDaMan

Yay NIST. Beautiful campus. Brainy people. 1 mile from my house. I may never leave.

So much of this is up for discussion still; the cloud is being thought of as a magic elixir for all things. The NSA thinks system administrators are going the way of the dodo bird... The robots will soon take over.

"NSA Plans to Eliminate System Administrators (August 9, 2013) In an effort to reduce the risk of information leaks, the US National Security Agency (NSA) plans to get rid of 90 percent of its contracted system administrator positions. NSA Director General Keith Alexander said that the agency plans to move to an automated cloud infrastructure..."

NSA to cut system administrators by 90 percent to limit data access - NBC News.com
NSA director
NSA gets burned by a sysadmin, decides to burn 90% of its sysadmins

pram

The primary benefits of the cloud are the on-demand and elastic nature of the service. A private cloud can be on-demand for sure, but not elastic. Having a private cloud does not magically result in more hardware or capacity. If you have spare capacity you are still paying for it. If you need spare capacity you will pay for it and wait for it to be shipped to your datacenter. Having Hyper-V or whatever behind a friendly portal does not change this fact.

pram

Really a Private Cloud even precludes the possibility of any significant horizontal scaling because of its inherent limitations.

colemic

LarryDaMan wrote: »

Yay NIST. Beautiful campus. Brainy people. 1 mile from my house. I may never leave.

So much of this is up for discussion still; the cloud is being thought of as a magic elixir for all things. The NSA thinks system administrators are going the way of the dodo bird... The robots will soon take over.

"NSA Plans to Eliminate System Administrators (August 9, 2013) In an effort to reduce the risk of information leaks, the US National Security Agency (NSA) plans to get rid of 90 percent of its contracted system administrator positions. NSA Director General Keith Alexander said that the agency plans to move to an automated cloud infrastructure..."

NSA to cut system administrators by 90 percent to limit data access - NBC News.com
NSA director
NSA gets burned by a sysadmin, decides to burn 90% of its sysadmins

We'll see how that goes when stuff starts to break or maintenance is due and they don't have resources to fix it.

'Sorry, that critical server needs to be rebuilt. I should be able to get to it in two weeks or so'

W Stewart

Before the change, "what we've done is we've put people in the loop of transferring data, securing networks and doing things that machines are probably better at doing," Alexander said.

Machine's are only good at getting exploited do to back doors left in by developers and just all around poor configuration on the part of the end user. Computers can only do things in routine fashion and they continue to work as long as things are running as intended. When things aren't running as intended, the computer isn't going to be able to fix itself. It takes a human mind with an understanding of the os, the hardware and software to figure out why it's not working and to rollback those updates when they break the system to the point where it won't even boot any more. A computer can't do anything but spit out repetitive and often cryptic error messages when **** hits the fan. These guys seriously undervalue the level of intelligence it takes to keep systems running smoothly.

apr911

Pram,

As a hater of the term "cloud" I do understand where you're coming from but I also think you are taking a very myopic view of the reality of what "cloud" is and frankly, your attitude towards it would seem to lend credence to the "adapt or die" theory as I think you may find yourself losing ground to peers who have accepted what cloud is. Before I get into your specific points of how Private Cloud is no different than Virtualization, I think its important to first understand how what we're discussing came to be about and why I so hate the term cloud so a bit of a history lesson.

Where did the "Cloud" come from?
The term Cloud originates from the networking world where the "internet" was often represented by a drawing of the cloud. Local administrators didnt know what what happened in the internet and frankly they didnt care since it was outside of their responsibility; they only needed to provide connecitivity to it and then let somebody else handle those systems.

But guess take that same drawing where all the clients and servers were nicely represented with the internet just being a generic cloud and invert it. Now your "local network" is a cloud and the internet is nicely represented... That's what ISP's did because they didnt care what their customer's internal networks looked like or how they were setup, it was outside of their responsibility; the only needed to provide connectivity to it and then let somebody else handle those systems.

And so a Cloud came to be the way to represent a black box system, something which you have no knowledge of the how it function but you dont need to know.

But then thats not the way the term "cloud" as we know it came about. So how did that happen?

Well, we as IT people failed to define a new emerging way of doing things that represented a massive shift in the way computing is handled. The problem is, since no one understood exactly what this new way of doing things was, we just decided to fall back to calling it the "cloud" somebody in marketing got a hold of that term and spun it into a marketing term and buzzword.

When we finally got around to defining what is today, we found not only had we incorrectly classified a lot of things as "cloud" (hence the invention of SaaS, PaaS, IaaS, NaaS and whatever other *aaS they come up with next) but we were unfortunately stuck with the term "cloud" as the blanket definition for it all.

So what is the cloud today?

Cloud today most often is used to describe software-as-a-service or infrastructure-as-a-service but more importantly, it describes how you access/interface with these services. Im not going to try to define SaaS Cloud as it is definitely the more fluid of the 2 definitions, to me isnt really much more than the next evolution and merger of software and websites but then since its so fluid, I could probably write another post just trying to define it.

IaaS takes a different spin on a 40-50 year old technology, Grid Computing, to make a large high-powered pool of computing resources out of numerous lower-powered systems and combines that with Virtualization; which is in itself a reincarnation with a different spin of an even older 60 year old technology/service in Mainframe Sharding (which is still quite profitable and in demand -- see Sungard).

Why I hate "cloud" as a term?

Cloud as a term is still very much a marketing buzzword, a way to take an old technology and make it new and exciting again but then so too is "virtualization" which while you seem so fond of it, I should remind you was not an entirely new concept and was merely the predecessor buzzword to cloud less than a decade ago. Ultimately though, despite the marketing spin, what cloud "is" hasnt really changed from what it "was" - a way to abstractly represent a system of which you have no knowledge of the interworkings but you know if you connect to it, it works and administering it is somebody elses responsibility.

There is still someone administering the cloud and they probably have a drawing where the users of the "cloud" are represented as a cloud and the actual systems that power the "cloud" are drawn out and interconnected.

So yes, I hate the term cloud and more specifically I hate the attempts to redefine cloud within IT since it is still just a way to represent an abstract system but the term seems to be here to stay sadly. To be clear, I hate the term cloud because I find it to be an asinine representation for what it actually is. If we had very early on defined what "it" is and gave it a term and hadnt let the marketing people latch on to something, Id be much happier but a virtual-sharded grid computing platform doesnt have the same ring to it as cloud or so the marketing folks tell me (personally I think VS-GCP sounds infinitely sexier than "cloud" but I guess thats why Im a geeky IT person and not in marketing).

apr911

So now that the history lesson is over, let get into some of your direct points.

RoberKaucher has already pointed out NIST's definition of cloud to which you argued:

On the definition of cloud computing:

pram wrote: »

We're not talking about the Cloud. Cloud and Private Cloud are not synonymous.

Sorry to break it to you but Public Cloud and Private Cloud are synonymous. The represent the same thing - a virtual-sharded grid computing platform. Public & Private are merely used to represent ownership.

On the definition of resource pooling:

pram wrote: »

Quite literally starting a server in Xen. You're using a shared iSCSI pool. Shared CPU. Delete the container and surprise those resources are available again.

You're actually quite right in this regard but since you want to argue cloud is just a different implementation of virtualization then Im going to argue virtualization is just a different implementation of mainframe sharding.

Cloud is the sum of all 3 technologies (sharding, grid computing and virtualization) which is why it has its own term.

On Rapid Elasticity:

pram wrote: »

How can a private cloud be elastic? Will each corporation simply purchase spare hardware? What happens if you exceed the hardware you have sitting in your datacenter? Are companies that are averse to large hardware purchases suddenly be willing to buy spare capacity simply because its now a Private Cloud? Doesn't sound very elastic IMO. If you shut down containers you're still paying for the hardware, electricity, etc etc.

On this point we couldnt disagree more. What you think just because its the Public Cloud that its infinitely elastic? Sure the big cloud players would like you to think that and to most companies and users it is (see old definition of cloud) but I assure there is a lot of capacity planning occurring behind the scenes.

In the private cloud, the cloud is still "infinitely" elastic as far as the users and company is concerned, but to the IT department you now have to do the capacity planning. No one, not even the big Public Cloud players, is sitting on a bunch of powered on, unutilized hardware. They capacity plan for 1, 3, 6 and 12 months out as best they can and adjust accordingly. The elasticity in the cloud isnt just measured by the ability to spin up instances but how well you can seamlessly integrate new hardware into or remove hardware from the existing pool.

If the cloud provider (either public or internal IT) finds they are sitting on hardware they dont need, they power it down to save costs. Yes, they did pay for the hardware and its now useless but there is going to be waste in any system. Public vs Private cloud only shifts the cost burden of that waste, however, a properly planned and executed private cloud can be run with better performance and cheaper costs than utilizing a Public cloud provider. There are pros and cons to both the public and private cloud which is why most people (at least in IT) wouldn't push for a pure public cloud play or pure private cloud play in their environment. Most likely some hybrid scenario would be adopted.

All of that is still just taking into account the computing side of the elasticity equation. What about financially? Sure virtualization has the ability to flatten the vertical cost line of purchasing new servers but as discussed, cloud is more than virtualizing on a single server.

No system is 100% elastic, they all have their breaking points but Private Cloud allows for the elastic utilization of exiting computing resources, the elastic expansion of the underlying resources and from an elastic cost that scales horizontally instead of vertically.

The private cloud does not absolve you from doing capacity planning accordingly, the only difference between Public and Private cloud is who is doing the capacity planning; you or some 3rd party.

Finally, On the topic of "If you are still doing things the same way as you did in the past, but just using virtualization, then you are by definition not doing private cloud."

pram wrote: »

No you're still using the same hardware as before. All you said is its going to have a portal to spin up containers. Xen has templates, everything has templates. No one needs to build machines unless you're running everything on bare metal lol.

Templates, containers and the like are great but are you truly giving your users on-demand access to these containers? From the sounds of things you do your capacity planning in the moment which means you probably dont. The user probably needs to request a VM be spun up for them at which point you review capacity, turn on the container and hand it over. How long that takes depends on you, not how long it takes to power on the container. This is not a Private Cloud.

I could honestly continue but the reality is I doubt I'd be successful in changing your mind. Perhaps one day you'll be forced to change your view or look back and go "damn, I should have gotten on board with cloud" or maybe you'll be lucky enough to continue to skate on by without having to do anything... Really it doesnt matter to me. As much as I hate "the cloud", Ive accepted this is the route we're going. The opportunities both Public and Private Clouds make available are changing the way companies approach IT. Even if the work we do behind the scenes remains fundamentally the same, it is having an impact on the way we approach our design & capacity planning of our networks and systems. You can adapt to those changes or find yourself with a skill set ill-suited to the realities of the work we do and watch your career wither and die.

NinjaBoy

petedude wrote: »

...Do you think this will see them start bailing out of local server platforms, and along with that certifications?

If they do get out of servers, I'd have to think open source platform certifications (e.g. RedHat, Cloudera) would start taking on more prominence.

What do y'all think?

I very much doubt that they will drop servers or their certification program, however they will evolve.

I mean recently I received an e-mail regarding a blueprinting opportunity for Office 365 (70-346, 70-347) which will lead to the MCSA Office 365. I can only imagine that they will be doing roughly the same thing for cloud based servers in the future, possible post "Windows 2016" basing on a new major release every 3 years...