Storagebod Rotating Header Image

Storage

Pillar-boxed (& buried?)

Jun 29th, 2011
by Martin Glassborow.
1 comment

I’d forgotten about Pillar in all the excitement of the various acquisitions of the past 12 months but it appears that Larry has managed to get the Oracle stock-holders to buy out his play storage start-up. Good work Larry!!

Does this have any real meaning for the rest of the enterprise storage world? I suspect not, things are moving on quickly and Pillar will simply be consigned to the annals of history as a rich man’s play thing. It’s probably more proof that Oracle really have little clue about the hardware side of things and should have stuck to software.

 

Sweet Violin Music

Jun 28th, 2011
by Martin Glassborow.
2 comments

Chris’ slightly breathless piece about Violin contains one quote from  Don Basile which actually almost justifies the whole breathless piece and he is talking about 3-bit MLC NAND

“The economic impact is that we’ll be down to $2/GB level pricing.”

At that sort of price, flash technologies start to come down to the cost of Enterprise disk in capacity terms and far out-strip Enterprise disk in cost per I/O. But the question seems to be will the traditional storage vendors embrace it?

There does seem to be a lot of negativity with regards to them embracing it but I am not sure that is warranted; EMC embraced SSDs quickly and the rest of the industry followed. And now most of the industry has technologies which allow them to make use of flash effectively; you are still going to be buying gobs of spinning rust for bulk data but if for instance you are EMC, you can build extremely effective and cost-efficient Isilon nodes filled full of flash and to hold both the meta-data and the active files.

In fact EMC’s Isilon purchase looks better every day; this sort of technology plays into the hands of anyone who has access to clustered, scale-out file-systems. EMC didn’t so they bought one.

And with Quantum’s recent announcement of a hardware appliance to support Stornext Meta-data; they also must have one eye with what is going on in the flash-space for bulk storage. I do wonder when NetApp are going to do the sensible thing and snap them up whilst they still can.

If you can buy flash at the price of 15K spinning rust, why wouldn’t you? It’ll take some years to get down to the price of SATA/NL-SAS but this is just the inevitable progression until the next great thing to come along.

Let the Campaign Begin!

Jun 23rd, 2011
by Martin Glassborow.
3 comments

So Zilla has fired the starting pistol for the race for EMC’s new CEO.

I would like to start the campaign to get Zilla appointed as EMC’s new CEO; he is the obvious choice and on that I think that we can all agree.

An EMC with Zilla in charge would be aggressive, ambitious and above all, amusing for us all to watch. He would be the Bond villain of the Storage Industry…planning for world domination from his underground base or at least his games room of doom!

Make EMC’s Court Jester the King!

Make EMC stand for Evil Mark’s Company!

The Call of the Cephalopod

Jun 21st, 2011
by Martin Glassborow.
No comments yet

As I continue in my experiments with scalable storage systems, I came across Ceph. Apart from having a very cool logo, Ceph is a distributed scalable network and storage file system which is based on an object store with a metadata management cluster which allows standard POSIX file-system interaction with the objects in the store.

As well as standard file-system’ access, Ceph also allows applications to interaction directly with the object store, an S3 compatible REST interface to allow applications designed to work with Amazon S3 to use Ceph as a private S3 Cloud and finally a network block device driver enabling Ceph to work as an ehthernet block-device like iSCSI; but the block device can be striped and replicated across the cluster allowing greater reliability and scalability (both capacity and performance).

Ceph is can currently be considered as alpha-code and you probably should not store your life work on it. That said it is relatively easy to set-up and play with. And as it released under the LGPL, you have full access to the source and can do with it what you will.

I built my cluster on top of Ubuntu and built it on three VMs; now, if I was building for performance, I would probably have ensured that I had built these on seperate spindles but I wasn’t and I was just seeing how quickly I could build it and get it working.

It probably took me a hour or so to build the VMs, install the packages,  configure the cluster and bring up a file-system. At that point, I just played with Ceph and treated it like a normal file-system, it works; writing a file on one node appears on all of the nodes in the cluster.

I’ve not had the chance to play with either the S3 or the native Object interface but I must say it all looks very promising.

There is an increasingly large amount of good work being done with scalable file-systems in the Open Source community but if you want to run a mixed cluster with Windows and Linux sharing block access to file-systems you are currently stuck with the commercial options such as Stornext and GPFS.

However, if you looking to build your own scalable NAS or Cloud; you have lots of options. And even if you still go with a commercial option, building your own is good place to learn what questions to ask.

 

Migration is a way of life

Jun 17th, 2011
by Martin Glassborow.
8 comments

As many of you are aware, my team is part of a major project to build-out a file-based workflow system for a major UK broadcaster. One of the things which is daunting is the sheer amount of data that we are beginning to ingest and the fact that we are currently looking at a ‘grow forever’ archive; everything we ingest, we will keep forever.

Even though we are less than two years into the project, we are already thinking about the next refresh of technology. And what is really daunting is that with our data growth; once we start refreshing, I suspect that we will never stop.

Not only will we be storing petabytes of new content every year; we will be moving even more old content between technologies every year. We are already looking at moving many hundreds of terabytes into the full production system without impacting operations and with little to no downtime.

We might be an edge case as we need to keep everything forever but as big data becomes a way of life for many companies; the migration effort will become considerable and the traditional migration techniques might not scale.

Data is going to have to flow automatically and between silos of storage technologies. Automated Storage Tiering was just the start; like it or not, the HSM problem is probably going to need to be tackled. And I can see some readers of my blog shuddering already!

 

 

 

 

 

 

Apple, XSAN and the Enterprise

Jun 15th, 2011
by Martin Glassborow.
4 comments

The bundling of the XSAN client with OSX Lion and the bundling of the XSAN Admin with OSX Lion Server basically acknowledges that Apple have now vacated the data centre space. If Apple had an enterprise class server hardware platform, there is no way that Quantum would have agreed to allow Apple to basically give away Stornext.

But it does give Quantum a trojan horse into the cluster file-system space; it will be very cheap for the home enthusiast to build their own scale-out block storage environment but if you really want to build at any scale, you are going to need start buying enterprise servers and then you’ll be into the realms of Stornext licensing costs.

I do find it amusing that Apple are happily bundling in Enterprise-like features such as XSAN, Fibre Channel support into what is at most an SMB Server whereas they continue to ignore technologies like iSCSI which fit nicely into an SMB environment.

And more interestingly, what Operating System is iCloud running on? What hardware? Is it OSX Server running on another manufacturer’s hardware platform? Or is it a completely different operating system?

It’s a bit like the stories that Microsoft for years ran their core accounting on something other than Windows. If you want to be a serious Enterprise player, you must drink your own champagne (which does sound something like Shirley MacLaine would do).

Baby VMAX?

Jun 10th, 2011
by Martin Glassborow.
3 comments

So Chris Mellor has used his journalistic nose to ferret out the VMAXe; so we have yet more overlap in EMC-land or do we? There has been a baby Symmetrix available from EMC for some time (DMX3-950 and DMX4-950 for example) and this is probably just a development of this policy.

I wouldn’t read too much into this move, I’m sure lots of EMC’s competitors will be jumping up and down claiming that this is a response to their product ‘x’ making inroads into the EMC’s core big-iron market.

I do find it interesting that according to Chris (and his sources in EMC are pretty good) that remote replication is Recoverpoint and not SRDF; this does however give EMC flexibility as to the replication targets and would allow you to replicate to a VNX for example.

For sometime customers have been looking for a standard remote replication technology within the EMC range but I suspect that EMC will support SRDF on the VMAXe as for many customers it would make sense as a development array or a recovery array.

I also like that it is factory pre-configured to be entirely VP; customers need pushing down this route, so far after some pushing, EMC included VP as a no cost option and now, they are factory pre-configuring. Does this mean that the VMAXe will not support the traditional Symm disk layout, I suspect not; in the same way that the VNXe tries to hide complexity but it is still there if you want it, the VMAXe will also try to hide complexity but if you insist, you can still make your life hard and complex.

I still believe quite strongly though  if you are interested in the VMAXe; you really ought to be looking extremely hard at the VNX (and its competitors) and if you currently have no Symmetrix on the floor at all, this is doubly so. But if you have mixed workloads and a requirement for mainframe connectivity (EMC aren’t supporting FiCON on the baby);  you won’t go far wrong with a VMAXe.

 

Simple Scales?

Jun 9th, 2011
by Martin Glassborow.
1 comment

One of the biggest problems that I and many others have with ‘Big Data’ is that it tends to imply that all data is the same; actually, that is probably a problem that the storage industry has been wrestling with for some time.

All data does not have the same value, life-cycle, access requirements, access characteristics but in many ways we have carried on as though it has. What is true with ‘Small Data’ is also true for ‘Big Data’; ‘Big Data’ comes in all sorts of shapes and sizes.

Does this mean that EMC are wrong to talk about ‘Big Data’; well, the answer to that is both yes and no. I prefer to think about ‘Data at Scale’; how do we manage all types of data at scale? And are we trying to manage data at the wrong level? Once it gets to the array, is it already too late to manage it?

It is interesting that what we are seeing is that when we start to manage at scale, we start to look at simplicity and stripping out much of the value add which the vendors have spent their development dollars adding to the traditional arrays. We instead look to the implementation of these value add features to the applications which are generating the data as it is at that level that the data characteristics should be understood.

The question for the storage vendors is will this trend continue? Is it simply because the features do not scale at present both administratively and technically? Or is management of data at scale simply different?

Is the era of general purpose storage coming to end? Will all storage trend to the lowest capability and rely on application intelligence to provide features? Or will the storage vendors find ways to clothe their products with new value propositions and complexity?

NDAs broken

Jun 8th, 2011
by Martin Glassborow.
6 comments

Okay, in an attempt to save the planet and stop people flying round the world to various briefings, conferences and junkets; I am going to bust open all the NDAs currently in place.

  • Storage controllers will be virtualised and the functionality will be abstracted into a portable software layer allowing us to use really cheap commodity hardware, refresh it often and stiff you massively for upgrade costs.
  • A common building block based on an open source operating system will be the heart of most storage controllers but we will charge you lots for something you could get for free.
  • SSDs will be supported with automated tiering capabilities and also with the ability to use them as a cache of some sort because we still can’t make up our mind which is the best approach.
  • FC backplanes will be deprecated and we are moving to a SAS backplane. We don’t have a real reason but it sounds cool.
  • Deduplication and compression will be integrated into all of our offerings. We bought a company and we need to do something with it.
  • Synchronous replication will transcend distance limitations and we will break the speed of light.
  • We will be more scalable than ever before and despite all the efficiency savings that you can gain from our features, you will be more profligate than you ever have.
  • We will tightly integrate with VMware and allow you to be even more profligate and provision 1000s of VMs that you do not need and do this extremely quickly. We also integrate with Hyper-V but you don’t care; the only Microsoft product you are interested in is Xbox anyway.
  • Our user-interface has been greatly simplified and even an idiot can use it. This is good because it will encourage you to employ idiots and idiots are easier to sell stuff to.
  • We have new strategies to address emerging markets including Big Data or as you might call it, somewhere to dump all the crap you’ve downloaded from the Internet and you don’t know what to do with.
  • We have a Object Storage product on the roadmap and will be introducing this to the market once we have a clue what it is.
  • We have an app….it’s cool! Look!
  • We have a Stack…..of Cash!

There you go folks, you can stop flying round the world or if you are, you can at least skip all the sessions and go on a decent bender in the safe knowledge that you are now fully briefed!

This briefing was brought to you courtesy of RSA and SecurID.

Scale Out Just a Feature?

Jun 7th, 2011
by Martin Glassborow.
2 comments

With Apple’s announcement that XSAN is now bundled as standard with OSX Lion Server; it appears that we are heading towards a world whether Scale Out Storage is just going to be a tick box on the feature list.

Although the announcement is fairly and squarely aimed at the Creative Markets,  we are seeing technologies from that vertical and the HPC vertical becoming more and more commonly utilised  as end-users in different verticals have to process more and larger data-sets.

Of course, XSAN is based on Stornext from Quantum; NetApp are Quantum partners, I’m sure this announcement has made the Apple fans in NetApp very happy.

We live in some interesting times.

← Earlier Posts Newer Posts →