Availability Advisor

New Uptime Appliance for critical virtualisation infrastructure management

Andy Bailey — Wed, 08 Feb 2012 14:45:39 +0000

In today’s consumer world one of the things we rely on are appliances – you know – white goods (although you can get them in different colours now) – cookers, washing machines and fridges. I always remember a question we were asked in one of my first physics lesson – if you switch the fridge on and leave the door open does the room get warmer or cooler?*

This has got absolutely nothing to do with this post by the way!

Anyway, back to appliances, there is a really neat one that I just have to wax lyrical about and it’s to do with the most critical part of your virtualisation infrastructure – it’s the bit that glues everything together and manages all those flexible resources and as such it must work all the time.

It’s so critical in fact that vmware came out with a heartbeat to kludge together a couple of standard servers.

There is a better way – see this!

*And, by the way, the answer is, the room gets warmer.

Do retailers understand the cost of downtime?

Andy Bailey — Mon, 06 Feb 2012 13:54:52 +0000

More anecdotal evidence that uptime matters.

We consumers have very high expectations these days. And it’s not just uptime either – quality of service (and response times) are paramount too.

Security Not the Only Risk of Virtualisation

Andy Bailey — Thu, 26 Jan 2012 08:24:28 +0000

We have all heard it – virtualisation is fantastic – it saves you lots of money and rewards you with flexibility. There are, of course, downsides …

One recoverable downside is security. This is covered admirably in this article.

Another, less publicised and equally recoverable downside is that you could be compromising your availability by hosting many different services on a single platform.

Easily recoverable! Just take out uptime assurance for your most critical applications and services. Talk to Stratus.

When Not to Virtualise?

Andy Bailey — Tue, 24 Jan 2012 08:06:12 +0000

Virtualisation technology has morphed from a mere enabler of efficiency into a fast-emerging bottom-line contributor. It is, in most companies, the default deployment option: you have to make a case for why not to virtualise, according to this article.

With the bank paying you a measly 3% (if you are lucky) it’s good to know that you can get a bigger return on your investment by virtualising. According to this article, 166%, 350% and 500% is on offer. Fantastic!

Just don’t forget though, that, if you put all those eggs in one basket, it needs to be a very strong basket.

Yes, I am talking about uptime here. If you want to assure your uptime, best talk to the experts. Either that, or don’t go down the virtualisation route …

How to Protect Virtualised Environments?

Andy Bailey — Fri, 20 Jan 2012 11:04:48 +0000

Here’s an interesting article from the Aberdeen Group about protecting virtualised environments.

Some sound advice here about tiering your applications to understand whether they should be deemed normal, highly available or critical.

This is especially true if you are virtualising them – and there is a note to say that over 50% of applications these days are deployed in a virtual environment.

The decision then becomes easy, as whether to deploy fault tolerant servers or make do with commodity.

Uptime Assurance for VMware Tier 1 applications

Andy Bailey — Thu, 12 Jan 2012 08:10:58 +0000

Had to mention this one now that vSphere 5 is available on a fault tolerant platform.

Just as the virtualisation stack gets better and better, so the expectation of availability increases too.

There really is no excuse for any downtime these days when it comes to the technology that is providing the backbone to those essential business operations.

It’s what keeps the revenues flowing.

How much availability is enough?

Andy Bailey — Tue, 10 Jan 2012 08:10:05 +0000

It’s something I have often mentioned and there is a good article here – it looks at “fail-through” systems rather than “failover” and how best to architect them.

So if you want the answer to how much availability is enough, have a read.

Hyper-Scalable, Cloud-Optimised Servers?

Andy Bailey — Tue, 27 Dec 2011 08:05:31 +0000

Hyper-scalable, cloud-optimised servers: why settle for less? asks the author of this article.

As good as commodity server hardware is these days, the fact remains that the architecture has inherent flaws in it.

What on earth am I talking about?

Ultimately, you have a single motherboard, memory subsystem and processor in there.

The more that manufacturers miniaturise this stuff, the worse the mtbf gets.

Do you really want to lose all that in-flight data that is sitting in memory waiting to be flushed out to the disks? Murphy’s law says that the system will fail just when it is mega busy.

You know what you have got to do. Protect your uptime with fault tolerance.

All I want for Christmas is a data furnace …

Andy Bailey — Thu, 22 Dec 2011 08:06:44 +0000

I think it is great when someone comes out with some whacky ideas that might just have merit. We all know that datacentres cost a fortune to cool and homes a fortune to heat.

It brings true meaning to the phrase “distributed computing” if it were possible to solve both problems at once by deploying DFs.

One thing is for sure, if you were to do this then you would need some ultra reliable servers to host said services. Those of a fault tolerant nature spring to mind!

Server Virtualisation – How to Pick the Right Model

Andy Bailey — Thu, 08 Dec 2011 08:32:27 +0000

Great article here, which heads off down the path of direct attached virtualisation vs. distributed virtualisation. I hadn’t really thought much about it before.

Anyway, what has it got to do with availability?

Well, firstly, there seems to be a huge difference in the quality of service that can be provided purely from a latency perspective. All good stuff.

Secondly, there is great debate about fault tolerant stuff and high availability. If you distribute, you are basically at the mercy of HA – and as we all know, this is a failure/restart scenario.

For true quality of service, make sure everything is on a server that is designed not to fail in the first place.