Exclusive It has only been a few days since HP decided to curtail its reselling agreement with Violin Memory to concentrate on its own 3PAR product. Big-mouthed bankers also had plenty to say on the move as it pertained to a rumoured IPO by Violin.
But El Reg has since heard a bit of inside information from various players close to the action that has shed a bit more light on the darkness of the HP and Violin relationship – and into Violin's selling and market efforts for its unconfirmed IPO. Here's what we were told.
An HP-Violin reseller relationship, background here, enabled HP to sell Violin Memory's PCIe-attached 3000 series products from July 2011. The idea was to combat Oracle's Exadata 2 system with an HP/Violin system that ran faster and better. You can see a 62-slide deck about this here (PDF). Slide 24 summarises the HP/Violin advantages over Exadata 2 in terms of $/IOPS, total IOPS, system cost and usable storage capacity.
We understand that this was a tactical exercise by the Business Critical Systems (BCS) part of HP, which is led by SVP Martin Fink (bio here (PDF), with the Violin 3000 products attached to several HP servers. BCS execs have told Violin this situation is unchanged, according to a previous statement by Violin.
BCS is part of David Donatelli's Enterprise Group in HP, called ESSN in the past, which has responsibility for industry standard servers (ISS); storage - run by David Scott; networking; and technology services.
A person within Violin said: "We never engaged with HP ISS or the storage folks," and added that they "don't know why they are being quoted (or being sourced) as there was never a deal with them. The 'non-news' seems to be sourced from a group that we never had an agreement with."
Of course, back when the Violin reselling deal was signed, HP had no in-house storage system that it could use to build an Exadata 2 killer. However, we are told: "There was always tension with HP Storageworks (3PAR) and we had various meetings where we agreed to 'just compete in the field' with 3PAR."
A source said: "HP put this product into BCS and the Storage sales teams just saw it as taking their business away. Because of this they just slagged it off wherever possible. Violin doesn't do replication with the HP version, so asking a client whether they were happy with no replication was pretty easy."
HP had nothing to say about this except to re-iterate that 3PAR is its strategic platform for flash going forward.
Fusion-io CEO and chairman David Flynn also mentioned the HP-Violin relationship in his company's quarterly earnings call, saying:
[The] relationship with Violin within HP has been strained for a longtime and we've been the beneficiary of that. They hadn't really made a very good partner, because their products are more in conflict with HP's existing portfolio.
And we take steps to make sure that our sales team is incented to work well with our partners. They actually make more when product comes through these partners. So there has always been a tension there within Violin.
Yes, this does present an opportunity, and not just at HP but within the channel and otherwise because our solution is an open solution that allows the customer the choice of their server. We're not selling a franken box, or some proprietary thing with a bunch of ... FPGAs and stuff, and that we're able to use [as] a Tier 1 server.
Violin also recruited Garry Veale, an HP guy in July 2011, when the reselling deal came into play, to be its European managing director. Veale was HP's EMEA storage VP and so ended up being the European face of a company that had sold its flash arrays into the BCS part of ESSN and also sold its flash arrays against ESSN's 3PAR and EVA arrays.
You can imagine that Veale was not popular in HP and this didn't do anything to enhance the position of the BCS-Violin relationship. In fact, an insider told us: "HP also have a hatred for Garry Veale at every level, so this was probably the biggest blocker. There is a two-year agreement with Violin and HP, so HP will give the official line to you, but underneath [it's] totally broke." There aren't fifty shades of grey in the channel; there things become black and white very quickly.
HP had no comment to make about this.
We're told that Violin is really struggling in EMEA, hitting only 10-15 per cent of its sales targets and that Violin's top EMEA executive can "put resellers off".
The pressure seems to be high; five major members of the company's 20 or so EMEA staff have left in the last two months. Recent sales wins include JP Morgan, Deutsche Bank and Bank of America, "all forced into the purchase so the IPO could list them as clients".
One overall view expressed to us is that Violin doesn't have a properly run channel programme, so anybody who wants to sell its products gets preferential pricing.
Violin has been offered the opportunity to comment on the various points mentioned here and has been unable to do so.
What we need is an HP Exadata killer
Back at HP, El Reg imagines that the BCS people still need Exadata killers and that helps explain why the reselling deal continues until August 2014, nearly two years' hence. By the time it runs out we envisage HP having its own in-house flash array technology, 3PAR technology, that can be integrated with its servers and take on Exadata 5 or whatever version Oracle is shipping then.
From Violin's point of view, El Reg understands that this does not much matter going forward: Violin told El Reg: "[The] Violin direct sales force has rapidly expanded and moved away from HP over three quarters ago to push our new 6000 flagship product. That is the focus of the company regarding our business planning."
The 6000s connect to servers via Fibre Channel, iSCSI, InfiniBand and also PCIe. We might imagine that the 3000 line will be enhanced but the 6000 is the flagship and will deliver, we suspect, the majority of Violin Memory's revenue bucks going forward.
What we might have seen here is the outward aspect of an internal HP Enterprise Group decision to concentrate on an in-house HP flash technology and move away from any reseller deal dependency. That's understandable and to be expected. A reselling deal is never as strategic as an OEM deal, and they come and they go. ®