What I want to do today is the first in what will no doubt be a random series of in-depth posts about some of the technologies, companies and products that we at Mediasmiths find interesting, and which we think are going to have a real impact on the media industry.

When we first started Mediasmiths, one of the values that we wanted the company to become known for was independence - freedom to help clients make the right technology choices, without feeling obligated to push them towards "partner" solutions. We've been quite successful at this, and have taken a number of clients now through our technology selection process leading to recommendations that have sometimes surprised both of us. We'll be talking about that approach no doubt, but I just wanted to set a context - things that we talk about on this blog are things that really interest us as individuals and as a company - any of our clients reading this should rest assured that we're not giving advance warning for what we're going to be trying to push down your throats..!  

And so to our first subject - I want today to talk about ObjectMatrix, and their product, the MatrixStore. MatrixStore is a disk based near-line archive/library solution targeted specifically at media organisations - and we happen to think it hits a number of sweet-spots that everyone is worrying about in the transition to tapeless. From a technology perspective, there are a number of interesting things that sets it apart:

• Designed to be a multi-node architecture out of the box - drop a copy of a file into one node, and it's automatically duplicated on as many others as you like
• Based on commodity hardware, recognised filesystems and so on, and it doesn't mangle your data as it's being stored - if anything bad were to happen, you could literally pull a drive out of one of these, mount it in a caddy and get your data back
• Treats files as objects not just as files - so stores metadata alongside the essence, and supports file search based on that metadata either through a UI or an SDK
• Policy based - allows you to define what gets replicated, what is read/write or read-only, how long data needs to be kept for and so on Allied to this, it has this nice little notion called "vaults" - logically separate areas on the store, letting you keep different data sets completely segregated (e.g. by client, by series), and apply different policies to it
• Nice little UI called Dropspot to manually get files in and out and search for them; a decent Java SDK; and pretty slick integration with Final Cut Server and CatDV
• You get pretty decent transfer bandwidth

Oh, and the price is pretty reasonable as well. Now, the technology is all well and good, but as we all know, it's not what it is, it's what you do with it that counts - and I have to say that when I first met the guys and looked at Matrix Store, I didn't get it at all. Firstly, I was coming with a mindset that said that disks are far too precious and expensive to use as an archive medium - the right place for an archive, even a short-term one, was LTO. Secondly, the demo I'd seen was of MatrixStore under Final Cut Server which coloured my thoughts as well - great though it is, the idea of FCS being used as a long-term project repository and therefore needing an archive underneath just didn't resonate at all.

That was at first. However, spending some time with the guys and thinking about some of the key challenges that Mediasmiths clients face led to something of a lightbulb moment. Whether MatrixStore or disk-based long-term archiving is the right way forward remains to be seen - the guys are developing some interesting TCO models which means it's definitely worth a look - but it's absolutely spot on for three of the main problems that our clients have with Tapeless workflows.

First of these is the problem with tapeless acquisition media - that once media is transferred on to the edit storage and the memory card is wiped for re-use, the material is gone forever. If you delete material off the edit storage by mistake, then there is no handy tape to refer back to - and once you get rid of a bunch of rushes to free up space on your expensive disk storage, they're gone for good - again, that shelf of tapes is a thing of the past.

Secondly, being able to shoot using tapeless media is giving us the same problem that we had in the original days of DV - production teams are generating a LOT of high-bitrate footage, which takes up a lot of space on high-performance edit SANs - and which quite often has to be moved off that edit environment and back on again at random intervals and at short notice in order to make room for more urgent jobs. This is particularly acute when considering how to deal with long-running series or ob-docs, where material can be collected months in advance of the edit.

Finally, we're finding that even those clients that are investing in enterprise-wide MAM-based archives are tending to treat them as posterity archives - the issue of libraries of re-usable material heavily used by local production teams, especially in returning or long-term series, isn't being thought through. At the moment, we're seeing a collection of "solutions" for these problems - none of which are particularly elegant, and all of which seem to involve unholy alliances between beige boxes with RAID arrays in them, LaCie disks replacing tapes on shelves, and LTO being used instead of video tapes. So, thinking about more elegant solutions led us straight to the capabilities of MatrixStore:

• Preserving rushes - drop them into a MatrixStore vault, and have them stored read-only, and automatically copied across the building or across town to a second node. I've heard the "big beige box of cheap disks" argument and I don't buy it. Apart from the fact that cheap disks ALWAYS fail when you REALLY don't need them to, it's just as easy to delete a file off the big beige box as it is to delete it off the edit SAN. Something a bit cleverer is required.
• Projects that you really do need, clogging the edit SAN - archive the projects off the SAN on to the much cheaper MatrixStore at a decent rate, and bring them back again when you need them - and have the search capabilities to find what you saved months ago. Again, you get the security of multiple copies, if you want.
• Local libraries - put those library shots on to the MatrixStore with the associated metadata, use the search capabilities to find what you need (OK - thumbnails, browse proxies and so on are things that need to be thought through), and bring them into the edit as you need to. Again MatrixStore stops you deleting your favourite shot of the M4 by mistake - and as things go, it's quite a bit quicker than looking up the library number, requesting the tape, spooling through it, and re-digitising what you need from it.

Clearly there are other sensible ways of solving all of the problems outlined above, and lest it be thought that I'm an unashamed fan of MatrixStore, it's worth saying that I still think that there are some areas to address - integration into one or more of the enterprise-strength MAM systems is one area to be looked at; getting an integration into an Avid environment which is as neat as the FCS integration is another; having a flight-case version that can be taken on shoots is a third; being able to have the system more media-aware and able to do things like retrieve media based on a timecode offsets would be a fourth; and finally working out how the MatrixStore search capabilities best fit into an environment where asset searches are generally taking place on metadata stored in a MAM rather than in a filesystem is definitely one that will need some thinking through.

I also remain a big fan of LTO - for a very stable backup and deep archive medium, I think it's great - and some of the things that IBM have done recently on simulating a filesystem on an LTO tape are interesting. That being said, our view is that MatrixStore is a neat product that solves a couple of real pressing problems in media, particularly for organisations going head-first into tapeless workflows - and it's something that a couple of our larger clients are looking at quite seriously. And it's also built and sold by some very decent guys, who are clearly very passionate about what they do.

Worth a look.