Tintri – another AI Ops storage advocate

A company new to me, with the slightly odd name of Tintri, came to my attention at a Technology Live! session in Munich recently. Tintri says that it is a “purveyor of AI-Powered Data Management Solutions, purpose-built for virtualized workloads”, which sounds like it is what we all need these days.

Tintri, however, has a bit more provenance than some of the later riders on the AI bandwagon. It was founded in 2008 by Dr Kieran Harty, who had been VP of Engineering at VMware. It hit some problems and went into Chapter 11 Bankruptcy (which is a protection thing rather than a failure thing), and was then acquired, in July 2018, by DDN, the largest privately-owned storage company in the world. Current Tintri employees say that the founder did things right – with the operating and file systems – because they are still using his basic architecture to build innovation on. Tintri even has “diehard customers” who run its first generation VM Stores, even though they are unsupported now, because they still work. This is sometimes an issue, as people relying on Tintri can sometimes forget that they have it, it just works.

Talking to them at Technology Live, they seemed to understand where AI could help, and its limitations. I think we need a hybrid approach, with humans still in the loop, and Tintri seemed to have similar views. I quite like “the Tao of Tintri”, which starts from the observation that the traditional storage approach – lock file, read it, write it back, release lock – was not designed to handle the dynamic needs of virtualisation, which wants to treat storage as just another random-access hard disk. In a virtualised world, the traditional approach was costly from both a CAPEX and OPEX point of view, and often gave rise to performance problems, causing virtualisation projects to fail. Tintri makes virtualised appliances optimised for virtualised environments. It lists its design goals (all of equal value) as Data Integrity; Availability; Simplicity; Performance; Scale; Automation; Analytics; Self-Service; VM/Application Awareness. Tintri is now hypervisor agnostic and now also supports hybrid environments, from on-premises to the Public Cloud.

So, what’s this with AI? Well, Mark Walsh (VP, EMEA Sales) says that Tintri will use it only when it delivers benefit to customers, not as a marketing ploy. He wants to remind people that Tintri is still here, even if it doesn’t have the sort of issues that would get it free press coverage, and it is still innovating. AI is fine, but is it life-changing? For Ops, very possibly, but elsewhere Tintri’s “management tool with storage” already delivers autonomous operations, real-time analytics, performance optimisation (it deals with the “noisy neighbours” issue), predictive analytics and immutable storage (so the ransomware guys can’t encrypt your data). Tintri claims significant gains in reducing admin overheads and TCO – which means, more immediately, that Ops gets to go home and be with their families. Nevertheless, I suspect really achieving this needs a fairly mature customer, one capable of actually utilising the goodies Tintri gives them. Perhaps Tintri could provide more consultancy to existing customers as to the extra value that they could be getting out of their Tintri. Similarly, Tintri doesn’t sell a ransomware solution but it has tools (analytics and immutable storage) that could be part of an effective ransomware solution. Is there a partnering opportunity here?

Mark drives a Tesla. He loves it – but he wouldn’t let it drive itself, yet anyway. He claims that self-driving data management is actually here now, up to a point, anyway. He talks about the “Pizza Box Challenge” – order a pizza, unbox your VM Store appliance and by the time the pizza arrives, your appliance is serving up data and you never have to look at it again, if you don’t want to (I’m guessing that the AI can’t open the box and plug in the wires yet – but, hey, robots). Personally, though, I’d still like to have a human in the loop, for the emergent situation no-one thought of and the AI wasn’t trained on. What AI can also do, is to significantly improve the user experience, for Ops, using natural language.

I commented that one would need “explainable AI” – one might like what the AI does initially but of one doesn’t know what it is doing, how can one trust that it is finding future optimisations? Tintri predictive analytics probably provides answers, but prospective Tintri customers often don’t believe some of the claims it makes. So Tintri makes “conditional sales” – if the claims aren’t justified for you, you can return the product in 30 days and get your money back. I like risk that is shared between vendor and customer.

Tintri is now active on social media, so check out it out on LinkedIn. It has reasonably wide global presence (UK and Ireland; Germany, Austria and Switzerland; South Africa and Middle East; Nordics; Benelux; Spain and Portugal) but is only just starting to move into France and Italy. It does particularly well in Public Sector, Health and Education and Service Provider verticals, where as much bang per buck as possible is needed. Tintri doesn’t sell direct, it is a channel-only company

People who use Tintri seem to like it, and even write songs about it (don’t ask) but those who don’t use it or don’t realise that they do, are often unaware of Tintri. It needs to address that, of course, but it would be worth doing a little digging, to find out about it, if you need storage for virtualised workloads, I think – especially as it now supports both Cloud and on-premises workloads.