SDI wars: WTF is software defined infrastructure?

The Software Defined Infrastructure (SDI) war is coming, and it will reshape the information technology landscape like nothing has since the invention of the PC itself. It consists of sub-wars, each important in their own right, but the game is bigger than any of them. We have just been through the worst of the storage wars. …

[steve 27]

Sounds too expensive... what you are basically talking about is public cloud consumed within a private space. It has been predicted that private clouds will fall of a cliff over the next couple of years. Why do all of this yourself when you can consume from a public provider ? niche workloads on-premise everything else off.

[Anonymous Coward]

I don't think so.

what you are basically talking about is public cloud consumed within a private space.

Err, no.

[Anonymous Coward]

Use a public provider?

Are you being serious????

1) The spooks, hackers and the likes of Google have much easier access to your data

2) You are totally reliant upon a functioning network between your premises and the cloud provider

3) You risk your business going under if the Cloud provider either goes under or has a significant outage.

Sure, put the public facing bits of your Biz out there in a cloud but betting your whole business on it is just plain crazy IMHO.

I wonder if you have tried to get business continuity insurance when your whole Biz is parked on/in a public cloud? If you have please explain the conditions they put on it. I am sure that a good number of readers here would be very interested.

I'm posting this as A/C because we are going through this whole morass at the moment.

[Anonymous Coward]

Re: Use a public provider?

I'm not advocating putting everything on public cloud but some of your points are relatively easily answered:

1) The public cloud vendors, or the big three at least, are now fully aware of how important data security is with ironclad contracts that they won't look at your data and enough encryption, sharding, sharks with laser beams etc. to protect it from outside spying.

2) Unless all you users are in the same building as your servers then you're already reliant on a functioning network provided by an external provider.

3) Unless you're a seriously profitable organisation or public sector then there's more chance of your business going bust than some of the public cloud vendors.

Public cloud can't directly replace on-premise IT but if you take it up a level does it have to? If you look at it from a real fundamental - what services do I need to run the business standpoint - you can find the real gaps and assess the risks and then it's up to the business to decide if those risks are worth running.

[Mike Pellatt]

Re: Use a public provider?

Just make sure you have your data and your apps in more than one place.

http://www.channelregister.co.uk/2013/02/07/2e2_data_centre_calamity/

[Anonymous Coward]

Re: Use a public provider?

if you're running 24*7*365 instances in the cloud you're probably spending more money than you would be if you had a well designed DCiB with a decent piece of orchestration kit (Opennebula for instance) backed with a decent toolchain for TDI.

[Fatman]

RE: Why do all of this yourself when you can consume from a public provider ?

I won't repeat the comments made by another poster regarding some of the drawbacks of The Cloud&reg.

But, I must ask, what School of Damagement did you graduate from?

Because if you accept the following as a 'reasonable' definition of Damagement, then I feel it may describe you to a Tee:

Damagement: A person in a position of authority in an entity who makes decisions outside of their training and/or expertise that have the potential to be extremely detrimental upon said entity; and does so, disregarding the opinions of those who "know better" just because the decision maker is in that position of authority.

You espouse jumping into the cloud, "putting all of your eggs into one basket". $DEITY, I would not want to be you when the sledgehammer shatters those eggs, and they end up on your face.

[Cookieninja]

Sounds nice, but ...

When does any of this stuff ever work as well as we expect? I still remember my parents talk about these computers that were going to take away everyone's jobs. The very things that have kept me gainfully employed for the whole of my adult life.

Infrastructure will get better, and more reliable, and it will allow us to do more things and process more data. However, it will *never* be pain sailing and it will always keep a large number of people gainfully employed keeping things ticking over.

[Rocket_Rabbit]

Re: Sounds nice, but ...

You see I'm not so sure. The ultimate goal of any company is to have as little staff as possible - they're a pain and costly.

So it becomes a simple game of numbers - it always is. You can have 95% of the effectiveness for 10% of the cost based on a 5 year ROI. Let's face it, what CEO/CFO/CIO wouldn't snap that up?!

For most SMEs, I can see a reduction to a few semi techie staff and a manager. Only for the larger companies will it make sense to perform some real fine tuning.

[Anonymous Coward]

Re: Sounds nice, but ...

This is an area where I am applying an awful lot of my active time at the moment, the notion of being able to provide everything from accurate costings per unit, just in time delivery for hardware, rapid provisioning of tested environments, change management, release management and the like, it's all a very exciting place to be working~~~

It's an interesting place to be. And being able to provide people that single pane of glass to see how much things are costing, have costed and will cost in real time, now that's the kind of shit that people like...

[Anonymous Coward]

Re: Sounds nice, but ...

We can do even better: We could make the hardware from "computing fabric" so we don't have "servers", instead we will have large gobs of FPGA-like units that are first loaded up with IP blocks in VHDL-like code, forming the "hardware", which is then customised by SDI blocks -> software all the way down!

[Anonymous Coward]

So instead of buying a f***ton of cheap proven storage,

Or even renting a f***ton of forrin storage from some shady rep you've never met,

You buy a f***ton of new expensive unproven hardware to juggle your data and drag it around between lots of people you've never met without you really knowing where any of it is?

Does this remind anyone of the financial industry, where the trick is to invent some entirely new financial 'instrument' (as they call it) that seems perfectly robust, shift as many as you can, then run off with the cash before anyone has time to actually think about it for a bit?

[M7S]

SDI - Suddenly I'm reminded of Ronald Reagan

And the various shades of Armgeddon anticipated in the 80's that meant HMG thought it would never have to make good on the state and public sector pension schemes.

Thanks for that. just in time for the weekend.

[Anonymous Coward]

Re: SDI - Suddenly I'm reminded of Ronald Reagan

Didn't you know? We ran out of TLAs in the mid-nineties. At least some of them are being recycled.

[Anonymous Coward]

Re: SDI - Suddenly I'm reminded of Ronald Reagan

I think the basic unit should be termed the ClusterF*ck rather than SDI. No doubt it will earn its name.

[Anonymous Coward]

so when SDI meets NFV

we'll have Software Defined Infrastructure meeting Network Functions Virtualization

I think http://forums.theregister.co.uk/forum/containing/2331420 was right!

[EssEll]

"All monitoring, backups, networking, storage configuration and so forth will simply take care of themselves"

HAHAHAHAHAHAHAHAHAHAHAHA

Someone's started early on the Friday beer goggles. This is the planning equivalent of "MAGIC HAPPENS HERE". It's a fantastic dream...

[Christian Berger]

It's a buzzword...

nothing more nothing less, just like all those SD* TLAs before, and, in a larger perspective, just the same as the many "shiny things" Microsoft promised in the early 1990s, like object oriented file systems, or Visual Basic for Applications which can incorporate your VBX (or later OCX) objects. Lots of shiny things we all stood in front of looking amazed. Today we start to see VBA as a pile of toxic waste and even Office warns of them by default. The object oriented file system was constantly promised for the version after the next version, and eventually got abandoned.

So in all seriousness, this doesn't seem like anything that's worth looking into. And even if parts of it turn out to be worthwhile, those should be easy enough to learn. After all all great things in IT are simply. Simplicity is necessary for something to really succeed and become a truly useful tool, instead of just a weight to keep you down.

[Anonymous Coward]

This sounds like the guff that lots of providers claim about their cloud solutions and that marketeers and analysts claim about the cloud in general.

Anything clever enough to do all of the above will either be super-expensive (either as custom hardware or software vendor licenses) or super-complex to develop and tweak for your own requirements.

Google appear to do something like this already, because they have a number of relatively simple applications that need to scale to the moon and so the problems are commercially worth solving.

Enterprises have the opposite problem. A number of smaller applications of varying criticality and complexity that a one size fits all solution will either be too expensive or too unreliable for.

Similarly, cloud options today come from the all-singing-all-dancing enterprise grade functionality out of the box sort (VMWare) to the lowest reliability commodity hardware with a highly sophisticated API set that lets you build reliability yourself if you have the time and resources to plan here (AWS and others).

Taking your VMWare specified solution and putting it like for like in AWS will save you a boatload of money, but lead to disaster when it doesn't fail over properly. Taking your AWS solution and putting it like for like in VMWare will drive you to bankruptcy. That's why enterprises with smaller apps and less time to spend fiddling trust VMWare and out of the box reliability and pay for it while web scale outfits go to AWS and invest heavily in building software layer resilience.

That was long-winded, but the key lesson is that this SDI article is nonsense. Some of the functionality will be useful for some people in some situations, but costly features that add no value will inevitably be weeded out to leave an array of different options for different types of client.

[ROIdude]

I would think that Cisco would have learned its lesson with WhipTail.

[Destroy All Monsters]

Did they forget the safe word?

[Anonymous Coward]

Software defined infrastructure......

....what could possibly go wrong?

I seem to recall an AWS failure......a new release of some infrastructure service could not be tested (because it required a test involving multiple data centres), so Amazon just deployed the new software. What are the chances that this vision has all the same characteristics....multiplied by 100 or a 1000?

[Anonymous Coward]

One big fail waiting to happen.

Sounds rosy, cost effective and focused on the C-Level.

However, the article mentions "security" as something to be delivered without saying *how*.

Something this important has to have security baked in from the start - layering it on top afterwards will be ineffective against attacks and nullify all the benefits of the automation and integration.

Disclaimer: I work for a security $_VENDOR...

[Trevor_Pott]

Re: One big fail waiting to happen.

If you'd like to discuss some of my thoughts on the incorporation of security, I'd be happy to have a sit down with you. As for why I didn't go into the details on the "how", it is because there are multiple possible approaches, and which approaches vendors take will differentiate them one from another.

It should also be noted that privacy should be baked in from the start...something not all vendors will do. (See: Microsoft.)

[The Morgan Doctrine]

SDIs Give Akamai Nightmares

Check out the billboard run in the Miami airport all October. Turns out, SDIs have obsoleted CDNs, the bread-and-butter cash cow of Akamai. As Akamai's customers arrived in Miami at the annual EDGE conference, a billboard greeted 100% of them. Red claw marks on a black wall, with nothing but the words "We Give Akamai Nightmares . COM" as text. Yes, there is a practical use for SDIs, and it's going to significantly change the balance of power in the delivery of Internet content. My question: How the heck did Mr. Potts "grok" this reality from publicly available information? IMHO, a brilliant analysis of technological data exhaust.

[dan1980]

"This means that spinning up a workload will automatically configure networking, firewalls, intrusion detection, application layer gateways, mirroring, load balancing, content distribution network registration, certificates and so forth."

There's no way that could ever possibly go dreadfully wrong . . .

"So you want a MySQL database tuned for the SDI block you are running? It will deploy a golden master from the orchestration software pre-configured and pre-tested to run optimally on that hardware. Your data and customizations are separate from the OS and the application itself. When the OS and app are updated, the image will be altered by the vendor; you simply restart the VM and you're good to go."

Sounds nice, but that's a lot of hand-waving - poof: "the image will be updated by the vendor", watch as my lovely assistant steps into this mysterious box - what will happen?

"Open the box, plug it in and you have compute, storage, networking, backups, disaster recovery, cloud gateway, WAN optimisation, monitoring, analytics, alerting and so forth all preconfigured and ready to go."

. . . which will, of course, meet all regulatory standards and specific needs of the business.

"When you move a workload onto an SDI block, it will detect the application(s) involved and automatically configure monitoring, alerting, backups and so forth . . ."

Again, sounds nice but how will it do this? All applications? What are we meaning by applications? VDI is an 'application' that runs 'applications' - will these magical boxes understand what is running inside the VDI sessions and will its backups and monitoring be aware of that level?

This is all a very nice idea, to be sure and I see it as a good goal to work towards. I appreciate that the article is intentionally light on the 'how' because there are many options and there will likely be a range of options.

One thing I see as potentially problematic is licensing, which may increase significantly when run on such infrastructure.

[jelo]

So we add more complexity, more software layers between keyboard and tin, more automation, and we'll have LESS administrators?

Very funny. I'm with Dan1980 on this. Not every database is the same. How in hell can you magically determine the optimum tuning for App XYZ? How does any auto bot decide what is online, nearline and offline data?

In fact, a home-grown browser-based ASP.NET app with diverse business rules: how do you magically determine the best config for a multi-tiered app with EDI interfaces between private, public and government entities?

Have you actually THOUGHT about what you write?

[Trevor_Pott]

The honest answer to that? Big Data. There are dozens of companies right now offering various cloud-based analytics software offerings that place an "observer" or "agent" in your datacenter. They then hoover up fucking everything. Every scrap of performance data. What's installed where. Peaks and valleys in response times for various infrastructure components, you name it. (See: Cloudphysics, amongst many, many others.)

Then you get into companies like VMTurbo that are now using this data to predict required changes and configurations...and they're getting quite good at it, even when they don't have access to Cloudphysics-like datasets.

Now, as a large company, you start buying these guys up. Not for the software they offer, but because they employ the best Big Data PhDs in the world, and they have amassed petabytes of data that is supremely useful for building out this level of automation. Your first generation robot handlers rely on statically collected information from volunteer canaries and non-automated deployments still using the cloudy analytics stuff. Not perfect, but that's okay, you're not automating the whole world yet; it's early days.

Meanwhile, the boffins are in the back room correlating application design and hardware design with various statistics and building models of how changes in applications will affect the results...then testing them. They are learning to build highly accurate predictive mechanisms that will make VMTurbo look like a child's toy.

And on and on it goes, getting ever more accurate. Instead of needing the "laying of hands" from the High Priests, this sort of stuff is dealt with by using empirical data, advanced prediction algorithms and high-reactivity monitoring that will catch any deviations from the predicted algorithms, adapt, feed that information back into the Big Data systems and refine the algorithms some more.

I should also point out that I've seen prototypes of this stuff actually working, and working on software and configurations never before seen by the prediction algorithms. I've seen them working on dynamic workloads. When you're a tech journalist, you get to see some of these stealth-mode startups. And then you start putting what they offer together with what these other guys offer, and you see that this company is making these acquisitions over here...

So..."how will all this black box magic voodoo work?" The same way a B2 Spirit Bomber stays in the air. Damned fine engineering. Modelling, modelling, modelling, and a fly-by-wire system that makes changes faster than any human could ever dream of doing.

You are about to become obsolete, sirrah. I know you won't believe that until it's upon you and you are staring at your own pink slip, but It's time to upskill.

Resizing LUNs does not add value to the business.

[CPU]

Cloud as a Bureau?

So all we need a is a dumb terminal to run things (plus TBOFH): the crunching, the software, the backups etc can all be done inside the cloud nebula. Sounds like the Bureau days will be returning :-)

[Anonymous Coward]

I'm curious to see if any one wants to go back and revise their comments in light of where things stand today.