Storagebod Rotating Header Image

Cloud

Blue Sky Thinking

Aug 28th, 2009
by Martin Glassborow.
3 comments

After yesterday's slight rant about the mis-use of the word 'Cloud' and everyone is using it to put in front of their product name and claiming what they sell is 'Cloud'; I thought I'd write a more reasonable piece on Cloud. Cloud Storage to be precise!

We often talk about an explosion in unstructured data and the huge amount of growth that is seen there; well, in my experience within the more traditional corporate IT functions, I don't really see it. Yes, there is growth but most of it is due to the mis-use of corporate file-servers to store personal files such as photos, music and movies. But there are only so many documents and work-related content that a single person can produce.

Obviously, in the media space; there is an explosion of unstructured data, content which would have been stored on traditional media is now moving over to digital media but most people do not work for large media companies.

But every individual in their personal lives are becoming content creators and they also consume an ever increasing amount of content. I talked very briefly about storing your life in this blog entry. It is this phenomena which will really drive the growth of storage going forward and it is this problem that needs to be solved.

Service providers are springing up to try to solve this problem and provide service to the end-user; almost every ISP tries to bundle a small amount of on-line storage for their customers. But to get a reasonable amount of online storage is still relatively expensive; to store the hundreds of gigabytes of content that some of us have at an online provider is still prohibitively expensive in general. Certainly if you want access to it in a form which is seamless to all the applications that you want to use, I couldn't for example move my iTunes library to Mozy and continue to access it via iTunes; I still need to keep my on-site copy.

If you think hundreds of gigabytes is an exageration; just look at the growth in personal HD-recorders such as the Flip and almost every camera. People are just going to upload the content and store it most of the time, it won't take them long to be heading towards a 100 gigabytes.

This storage needs to be cheap and online; I am for the moment ignoring the bandwidth issues that many of us face today; this storage is going to have compete with the cost of DAS today and I am talking about consumer DAS; not enterprise drives.

This is the real problem that what I think of as Cloud Storage needs to solve, not the traditional corporate IT storage issues. Not the IaaS issues which are actually relatively easy to fix; all that needs is cheaper than the current Enterprise storage with an easy to use front-end. That is not a hard problem to fix and that is what annoys me most about some of the announcement and pronouncements that have been made about Cloud recently.

Let's start thinking about how we solve those things which are hard, not sitting in our little comfort-zone supplying the same old storage with go faster stripes. Because if the traditional vendors don't; there are plenty of non-traditional vendors who are not tied by legacy thinking who will.

Some of you have already made a start, some of you are pretending. This is a different game to the one you were playing in. The old game is still there but there's a new one to play as well. And what is Consumer today may become Enterprise tomorrow; this is not just a case of trickle-down, there will be significant cross-fertilisation.

There is a big prize to be won but I'm not sure it is the prize that most of you are playing for.

Trough of Despondency

Aug 27th, 2009
by Martin Glassborow.
No comments yet

Cloud must be going though the most rapid hype cycle in history; I now am firmly in the trough of despondency! Every hosting company is claiming to do Cloud; every supplier who supplies a hosting company claims every product that they sell to them is Cloud!

Okay, I am firmly in the camp that commodity-based IT infrastructure is generally a good thing. This infrastructure should be easy to provision, easy to scale, easy to support and have a financial model which means that in general, one pays for consumption as opposed to potential consumption.

From the end-user point of view, IT infrastructure should be invisible and as magical as electricity, the infernal combustion engine, powered flight and all other technologies which are used on a day-to-day basis and the average tabloid reader neither knows or cares how they work.

There are many technologies both old and new which enable this process; none of these technologies are in and of themselves Cloud Technologies.

The rise of the x86 Hypervisor as exemplified by VMWare is an extremely important enabling technology; especially if you wish to segregate workloads, either for technical reasons or to enable multi-tenancy type environments. But an environment utilsing a hypervisor is not necessarily a Cloud and a non-hypervisored environment is not disqualified as a being Cloud. 

Actually, I think the constant focus on the Cloud may actually be harmful; we need to be focussed on what the customer wants; if we spend our lives trying to shoe-horn everything into a Cloud-type definition, we could well miss what is important. And if the customer at the end of the day, nods and says 'Hey, you mean, just like a Cloud; you can smile and say 'Exactly!''.

So just because you interface well with the variety of Hypervisors out there; it doesn't mean you have a Cloud product. You are quite possibly doing yourself immense harm and you may find you alienate those people who don't want a Cloud product!

Simon Wardley from Canonical likes to compare Cloud Computing to the Industrial Revolution; Cloud Computing is a movement, a change; it is not a thing.

Stephenson launching 'The Rocket' as 'Industrial Revolution'!?

Duke Nukem Forever (OnTap 8)

Aug 25th, 2009
by Martin Glassborow.
7 comments

Unlike DNF, OnTap 8 actually looks like it is going to ship! This has been a long road and it is unfair to say that at various points it looked like that the version number was actually denoting the years it had been in development. But that's what it felt like!

The release of OnTap 8 is interesting; it's a big deal for NetApp and is almost certainly as a big a deal for them as V-MAX was for EMC. But what is important is that it is not currently linked to any new hardware platform; it is a software release. Yes, there are some hardware announcements but no major refresh.

This is currently a hugely important difference between NetApp and many of it's rivals; NetApp are a software company which happen to sell a hardware platform to run their software on. Their hardware platform is not especially special (sorry guys but it isn't); the special sauce is the software. And with OnTap 8, they have completely revved the recipe!

The feature list is impressive and contains many things which I have wanted from them for some time. Lots of good stuff around scalability; some of it long overdue, especially the enhancements delivered by 64 bit goodness.

I especially like the announcement around Data Motion; seamless migration between storage devices, a big deal in the NAS space. Moving NAS data around without outages is painful. If NetApp achieve this, more power to their elbow. Data Mobility is one of those problems which has yet to be fully solved.

I am not entirely convinced by the PAM-II cards and it is going to be interesting to see how the competing approaches to flash stack up long term.

So we've had V-MAX from EMC and OnTap 8 from NetApp. So what have HDS and IBM left to offer us?

Pink Fluffy Clouds!

Aug 25th, 2009
by Martin Glassborow.
No comments yet

Sometimes I wonder if we don't do ourselves a massive dis-services in IT; for example, take cloud. We keep hearing comments like IT departments aren't ready for it, they don't have the processes, they aren't flexible enough etc, etc..but I wonder how true that is?

Many IT departments have been through numerous re-organisations; re-assessing their processes on a regular basis and at least in theory streamlining their processes. We've been push-button provisioning servers for years pushing standard installs to machines with automated customisation tools. It would not have taken a huge amount to take these automated build scripts and put a web-interface in front of them; in fact, I know plenty of places who have. Just not told their users about it.

The on demand world has been a reality for quite some-time with much of the latency artificially introduced by ourselves. You see, I think that we are ready to do this and instead of waiting for vendors to come in to tell us that we aren't, we should be telling people that we are and just bring it on.

But I suspect it's both a political thing and a belief thing. There's some explaining to be done; how do you explain that you've been able to do something for years but just haven't? And how do you get your users to believe you now? In order to avoid embarrassment, we've just been offering incremental changes in OLAs and SLAs but how do we deliver the step-change which we are capable of delivering today.

Wrap it up in a big fluffy bundle and call it Cloud! Pretend it's something new! That way no-one has to be embarassed but I suggest that most cloud initiatives ought be coloured with a slight pink tinge!

Driven by Past Policy

Aug 24th, 2009
by Martin Glassborow.
2 comments

Every now and then, I like to annoy people and point out that much that we are talking about as the future in Open Systems has been done before. And today is one of those days!

I was talking to someone about policy driven storage management and what they described sounded awfully familiar, so I thought I'd do some quick googling and point them in the right direction. I found this series of articles on the IBM website.

It's worth reading up on System-Managed-Storage; there are so many concepts within it which should strike a chord with you especially if you are thinking about what storage is going to look like in the future. And probably like a lot of Bods, you've never had the chance to be indoctrinated into 'the Cult of the Mainframe'.

Sanitation

Aug 19th, 2009
by Martin Glassborow.
4 comments
As we build ever larger storage estates and generate ever increasing amounts of data; I realised yesterday that a project that I am currently involved planning of the implementation has no migration path out of. The decision that we have made is pretty much final, this is it; once this ball starts rolling, it doesn't stop and once it's built; it'll take the rest of my career to migrate out of.

And boy is that a scarey thought; this is a bet your house moment! I'm wondering how may people are actually heading down that route, making decisions which are going to have to be lived with for a lifetime.

I think that we've spent 40-50 years building open-sewers and I think we are going to spend the next 5 years building a sewer-system which will stand the test of time and will have to because digging them all up again is just going to be too hard. 

Never Forget

Aug 18th, 2009
by Martin Glassborow.
No comments yet

After posting yesterday on the Information Haze; I came across these guys who work for Microsoft and and have a book coming out in September.

This sort of project is a fascinating glimpse as to what is coming in the future and the implications it has for the storage requirements in the future; if you read their 2006 paper, they estimate to store an 83 year life it will take about 200,000 GB so 200 Tb. And that is conservative I suspect.

But more interesting is both impact it has had on them and I found it especially interesting the feelings of loss which were generated by the loss of a hard-drive resulting in 4 months worth of visited web-pages being deleted.

If you are saving your life to the cloud; I guess, you better have a back-up.

Information Haze

Aug 17th, 2009
by Martin Glassborow.
3 comments

Chuck riffs off Zilla to talk about a personal cloud and what it may look like; at the moment so much of our personal informaton is dispersed across a variety of places, we do not have a cloud it's more like a mist or a general haze.

It's not just content such as photos, mp3s, blogs etc; it's also meta-data such as Amazon wish-lists, eBay purchase records, grocery shopping lists, my Linked-In contacts…information which describes our everyday lives, what we do and who we are.

I am in the process of scanning in the barcodes of all the books in our personal library and uploading them to LibraryThing; once those are in, you will probably be able to tell quite a lot about me. It might tell you more about me than my medical records but I probably don't mind you looking at what books I read but I would really object to my medical records becoming part of the public record.

What I need is a tool to help me manage all this information about myself. Some of which I am happy to be public knowledge, some which should always remain private.

Personal Information Management is a massively untapped area; how do we define the taxonomy of a life? And who are we going to trust to store this information, how are we going to ensure portability? These are complex things and I wonder who is going to solve them.

Cloud==Convergence==Service

Aug 11th, 2009
by Martin Glassborow.
No comments yet

I've had a few discussions recently with people about Storage Teams, how they are organised and what perhaps they should look like in the future. There is a feeling amongst many people who sit outside the storage arena that storage is a deep black art; the successor to the deep black art that was mainframe system programming. And for many years, the storage teams and the vendors have colluded in this.

Ask a storage manager to do something for you and there will often be a pursing of lips, an intake of breath and a comment, 'well I could give you that but that's not really the way it works'; or a narrowing of the eyes and a shaking of the head, 'Hmmm….complicated, it'll cost you!'.

Indeed, the storage monkey is the plumber of the infrastructure world! Or at least this is what it feels like looking in but this attitude/position is beginning to be challenged more and more; virtualisation has meant that generalists are becoming more common and infrastructure ownership is becoming blurred.

Who is the storage team when a Lefthand software appliance is deployed? Who is the storage team when a Sun 7000 gets deployed? In fact, arguably any NAS product has more in common with servers than traditional block storage arrays.

Who owns the converged network? Much of the resistance to iSCSI came from within the storage teams aided and abetted by the traditional vendors. However we now have FCoE and we now have the first commercially available FCoE array; so we are going to have to sort out the politics and the state of denial that we've all been in because FCoE is fibre channel and we can't get away with claiming anything different.

There will be a converged network and that network will be Ethernet; we can have two teams attempt to manage it or we can have one. I suspect that many will try the two teams approach, just as when I was first involved in deploying an IP network that we kept a seperate team to manage the SNA network. But eventually, the teams will merge and become one.

But I for one look forward to this; I look forward to the time where the storage team isn't begging the server team for root-level access to do something because they should all be one team and they should be pulling together to provide Service; not servers, storage and network but Service.

And I think that is the biggest win that the Cloud could give us. Technically we have been ready for the Cloud for some time; yes, the management tools are a bit lacking but we are ready. Politically and emotionally we need to move out of our infrastructure silos and move on.

This will causes some huge problems for some vendors in the future; it is going to be very hard to be just a storage vendor, just a server vendor or just a network vendor. I think you can get away with that for another two-three years but long-term, I think you might struggle.

Tape No More

Aug 11th, 2009
by Martin Glassborow.
15 comments

I was on a course last week taking a trip down memory lane and re-learning all I knew about TSM; life has strange twists and turns but it made me think about back, recovery and tape.

Every year we have the conversation about tape and whether tape still has a place and I think I am now personally comfortable with answering 'Tape should no-longer have a place in backup and recovery!'

What has really changed my mind is the increasing growth of cloud-storage options which allow bulk-storage at a reasonable cost off-site; until recently, tape fulfilled this role but I think that we can do this now all in the cloud. I can think of no reason for most small businesses to do anything else; products like Mozy can support them and there is no longer any excuse for not having a robust back-up regime.

And the traditional back-up tools will need to start including options to back-up and restore from the Cloud.

← Earlier Posts Newer Posts →