A slight adjustment to Njini's knitting
Becoming a fan of network file management
Njini used to say that Information Lifecycle Management (ILM) was flawed because it didn't treat information as an asset; it's basic goal, when espoused by hardware vendors, was to sell more hardware. The ILM goal was good but you had to properly organise your files of unstructured data, growing like topsy, before you could do any sensible ILM.
Its IAM product was just the software you needed to do this. That was 12 to 18 months ago and the ILM marketing fires are dying down, being replaced by file area networks (FAN), disk-to-disk backup concepts like virtual tape libraries and de-duplication, and energy costs of data centres.
Naturally Njini has adjusted its messaging and is now onboard the FAN and datacentre energy cost trains. This became apparent during a discussion with Njini CEO David Jones. It was meant to be an interview but the gently but firmly persistent David Jones, turned it, in the nicest possible way, into a series of mini-lectures.
Like a gentler version of Arnie's Terminator, David Jones won't be diverted by questions. He just keeps coming at you and will never, ever stop until he's put his points across.
Steroidal iPods and vectors
"We have around a dozen customers. We're selling to companies with 3 - 10TB of storage, 1,000 - 2,000 staff. That's our sweet spot. They struggle with the same problems as corporates but have fewer resources. Existing storage is like an iPod on steroids. Basically it's software that's the value. Potential customers are struggling with three challenges. They have the tyranny of vendor lock-in. Second the rate of new file creation exceeds their ability to provision storage and, thirdly, there is the change in the regulatory environment."
"Storage is a sleepy backwater. There used to be one answer to any storage problem; sell more disk. It's what EMC and NetApp are geared to do. But it doesn't respond to the three problems above. These are information management problems, not storage problems. That's why we don't face competition from NetApp and EMC. They don't support single instancinmg of files because it reduces the need for disk. (But see the note below on EMC and SIS.)"
"There are two vectors for storage: how do we reduce storage cost; and how do we protect data. Little markets have grown up around point solutions to aspects of these problems but the overall area is too complex."
"ILM is a strategy to reduce the cost of storage. But why hasn't it taken off? It tends to be offered by storage vendors and work with their own hardware. It needs to be agnostic though. The tool it's based around is a file's last access time, used to predict the rate of decline in activity for a file. It's just not granular enough. It's fundamentally wrong. How can you write policies around it?"
"We talk about it as network file control providing policies and a single point of control for the two vectors. A really big motivator is to reduce the cost of storage. Enterprise storage is a $15 billion market. There is an easy 20 - 30 percent shrink in storage cost by single instancing - de-duping at the file level. That means taking $3 billion out of the market. Our value proposition is drop dead simple."
"Other suppliers claim to de-dupe but the hard thing is file change after a de-dupe. To solve that problem you have to be in-line. Njini saves changed files as unique files. Now people can implement ILM because we give you granular control, all JPEGS go to tier two storage for example.
May the enforcement be with you
"Concerning compliance we can help with regulatory oversight by enforcing access policies to files, specifying access rights - who can read, who can write to a file, who can delete. To provide enforcement you have to sit in the data path. This is very different from Google Enterprise, Kazeon and Scentric (partnered by Hitachi Data Systems). They do it after the fact, once files are on back-end storage."
"Because Njini's IAM product is inline we are uniquely positioned for network file control."
A fan of FANs
"The market is moving towards File Area Networks. We'll need to work with the file virtualisation vendors. It's key. It's driving us towards OEM-type relationships."
"Njini is a member of the FAN working group set up by Brad O'Neil of the Taneja Group. Other members include EMC, Brocade, HP, Microsoft and Acopia. The aim is to produce a deliverable, a set of neatly-defined interfaces, and take it to SNIA. The first meeting is on December 19th and will be hosted by Brocade in San Jose."
A first rule in marketing a company and product is to stake out unique territory in the market, a competition-free zone. A second one is to adapt to events. Njini is being very ingenious at this, positioning itself as the only on-line, information (file) classifier, search-provider, access manager and single-instance storage provider. As a result of this positioning the other players in the file classification, FAN game are postioned as being complementary to Njini's product or out-classed because they are not, for example, in-line and hence unresponsive.
Jones said: "From a functional standpoint there is no one we compete with. There's certainly not competition from the off-line classifiers: Kazeon; Scentric; StoredIQ - that's search basically. We do asset policies."
Njini has had a third round of funding and part of the cash will be used to build its sales and marketing infrastructure in Europe. A fourth round is anticipated in 2007 when the company might move into profitability, with a potential IPO in 2009. Jones describes Njini potential this way: Njini is a really nice investment. It's a billion dollar company or it's nothing. Remember Veritas; it rode backup to a $10 billion valuation."
EMC and single instance storage
In point of fact EMC is going way beyond single instancing of files and promising potential customers an up to 300:1 reduction in disk capacity needed to store backed up files through its Avamar acquisition.