This is the second entry in my Top-Ten in Storage series.
Not every innovative product can succeed in the market, and no matter how good some ideas seem, they can fail to make much of an impact. The truth is, people buy solutions, not technologies.
This list includes products so cool, so ahead of their time, that they just couldn’t fail. But they did.
1988 EMC Orion/Allegro
EMC’s RAM-based storage array predated and presaged its Symmetrix, offering lightning-quick I/O for very small workloads. But the Orion had a few serious drawbacks: It was astonishingly expensive, offered tiny capacity, and was never available for open systems. EMC’s initial offerings in the open systems market, Centriplex and Harmonix, also deserve mention though the company proved its worth with its “best” Symmetrix a few years later.
1985 3COM 3Server
I’ve written about the 3Server before, but let’s sum up: Before RAID, SAN, and NAS, and even before NetWare and NFS became entrenched, 3COM introduced a flexible multi-disk network storage and print server called the 3Server. It was an innovative combination of hardware and software, but it lost out to NetWare on open PC hardware in the market. Although the 3Server lasted just a few years, it made a lasting contribution to the field: Its 3+Share software became 3+Open, then LAN Manager, and was finally absorbed by Microsoft to become SMB/CIFS.
HP AutoRAID (High Availability Disk Array Model 12)
The first post-RAID virtual sprang fully-formed from the big brains at HP to challenge the industry-leading modular offerings from Data General’s CLARiiON and DEC/Compaq’s StorageWorks. It one-upped everyone with automatic relocation of LUNs between different RAID levels based on workload about a decade before Compellent, EqualLogic, and 3PAR had anything of the sort. But the AutoRAID had two key drawbacks: Like mama bear, it wasn’t big enough for big companies but was too big and expensive for little ones, and it was SCSI-only (though a Fibre Channel bridge was a common accessory). Some of the concepts lived on in the HP EVA line, but AutoRAID didn’t last long.
Microsoft Object File System (OFS)/Relational File System (RFS)/WinFS
Microsoft shoulda-coulda-woulda revolutionized file storage with nearly every major operating system release. It looks the same every time: A database/filesystem hybrid that would add structure to the vast unstructured file world. Despite the hype, we haven’t seen it yet.
MangoSoft Medley
Consider the typical LAN: Lots of clients with lots of hard drive space connected to a server with yet more disk capacity. Wouldn’t it be nice if the space on these clients could be used as a virtual network server? Well that’s just what MangoSoft announced in 1997, gaining much attention in the PC press. Sadly, the idea never really caught on, although MangoSoft continued. But distributed network storage has become the industry’s Brigadoon, trotted out as a new idea year after year.
StorageNetworks
This is my flop – I worked for StorageNetworks from 1999 through 2001.
StorageNetworks was launched in 1998 to provide off-premise storage and backup as an on-demand service for enterprise customers. They sought to take advantage of the emergence of high-speed Fibre Channel connectivity over metro distances to commoditize storage capacity. But only Houston (where I worked) had an acceptable infrastructure for the service, so the company changed focus to hosting centers. Then off-site backup. Then software. Nothing worked apart from vendor-independent services, and the (now public) company was shy to base its revenues on that. StorageNetworks was gone in 2003, but does the service remind anyone of Amazon S3 or Nirvanix? Maybe it was just ahead of its time.
Compaq VersaStor
Announced in 1999, VersaStor would have been a revolution in Fibre Channel SANs, a full out of band virtualization solution leveraging specialized HBAs directing traffic. It was continually pushed back, finally being “merged” (after Compaq) with HP’s StorageApps in-band SANLink to become CASA. But then EMC sued HP over virtualization patents, derailing CASA, and the whole mess was permanently shelved at the end of 2003.
IBM Ice Cube/Collective Intelligent Bricks Hardware
IBM rocked the storage press in 2003 with their announcement of Almaden Research’s Ice Cube concept. Instantly dubbed “Lego brick storage” (trademarks be damned!), the storage units could be stacked in two dimensions, scaling without limit. Coolest of all (literally), the bricks were chilled with water! Although the concept progressed, we still haven’t seen it. But this didn’t stop Seagate from developing a similar concept, ISE, which is now Xiotech’s main offering. IBM’s Storage Tank and VSS could have been nominated, too, but I’m not a sadist.
Revivio CDP
Revivio was the pioneer of continuous data protection (CDP), with great technology and people. But no one (other than Symantec, eventually) bought it.
Pirus
Little Pirus was working on a small but scalable virtualized target when it was acquired by Sun in September of 2002. Their technology was launched as the StorEdge 6920, a mini alternative to the HDS USP in 2004, but (according to insiders) it didn’t exactly light the world on fire. Sun finally pulled the plug on the 6920 in early 2007, with HDS taking over continuing support for anyone who bought the moribund product.
Stephen how about the Memorex Solid State Device (SSD) that thankfully most of that generation has or is about to retire removing the negative stigma of SSD they experienced. How about the Britton Lee database machine (Hey Larry, you might want to research why it failed 20 years ago), or StorageTek SN6000 (ok, Im going to get some flack on that one), or, the Encore sold to Sun and sold as the A7000 Symetrix wanna be for mainframe attachment, or, MTI (formerly SF2) Failsafe although given the $$$ MTI made off of it, tough to say if the balances the reputation spawned by DEC (e.g. now HP) as the Failsome, or not to be outdone, DEC and later Compaqs desire to see 1394 Firewire as a universal I/O connectivty for large systems in place of SCSI or Fibre Channel (hummm, I wonder if we will see a 1394 version of Xiotech ISE 😉 ), or what ever happended with RISS which many people commonly mistake the HP MAS (bycast) as being.
In addition to SNI which as you pointed out which later re-morphed into a couple of other companies (e.g. Gianloop and then Centrepath) to try and sell the management and management software layer and is still trying from last I heard, there were also all of the others including Storageway, Storability (morephed into SRM and sold to STK), Space4rent, etc. etc. etc. etc.
On your list Stephen, software is getting off pretty easy as there has been some serious $$$ droped in that space with many now on the where are they now list (or soon to be) from the SRM, eDiscovery/Search, Database tools related among others.
Perhaps that can be your next list Stephen, where are they now, many of which have become road kill, or their IP bought up for pennies on the dollar. Here’s a few to get you started:
Creekpath – Finally sold after how many millions invested
Crosswalk – Jack McDonald (aka founder of McData) SRM company that redefined the 180 degree change in direciton by becoming a clustered storage system solution who then ceased operations
Crosstor – Sold to EMC, some of the IP still lingers around
Giganet – RDMA type Ethernet based technology bought by Emulex
Highground – What did Sun ever do with it?
Imperial – One of the original SSD vendors that went throuh a few lives
Intradyn – The SMB focused email and compliance appliance reportedly recently sold at a bargin (and not to Sony)
MTI – Should have been left on Autopilot when Earl retired
Neoscale – Security and encryption assets sold to nCipher
Neustream – Open sourced based multi-protocol storage ceased operations
Sancastle – Im sure Marc Farly can tell us about that one
Sandial – IP was purchased after operatoins ceased
Stonefly – Now DNF remorphed into a storage solution vendor
Ouch, some of these are bringing back a headache, oh well, have fun and see what else comes to mind.
Cheers
gs
Stephen how about the Memorex Solid State Device (SSD) that thankfully most of that generation has or is about to retire removing the negative stigma of SSD they experienced. How about the Britton Lee database machine (Hey Larry, you might want to research why it failed 20 years ago), or StorageTek SN6000 (ok, Im going to get some flack on that one), or, the Encore sold to Sun and sold as the A7000 Symetrix wanna be for mainframe attachment, or, MTI (formerly SF2) Failsafe although given the $$$ MTI made off of it, tough to say if the balances the reputation spawned by DEC (e.g. now HP) as the Failsome, or not to be outdone, DEC and later Compaqs desire to see 1394 Firewire as a universal I/O connectivty for large systems in place of SCSI or Fibre Channel (hummm, I wonder if we will see a 1394 version of Xiotech ISE 😉 ), or what ever happended with RISS which many people commonly mistake the HP MAS (bycast) as being.
In addition to SNI which as you pointed out which later re-morphed into a couple of other companies (e.g. Gianloop and then Centrepath) to try and sell the management and management software layer and is still trying from last I heard, there were also all of the others including Storageway, Storability (morephed into SRM and sold to STK), Space4rent, etc. etc. etc. etc.
On your list Stephen, software is getting off pretty easy as there has been some serious $$$ droped in that space with many now on the where are they now list (or soon to be) from the SRM, eDiscovery/Search, Database tools related among others.
Perhaps that can be your next list Stephen, where are they now, many of which have become road kill, or their IP bought up for pennies on the dollar. Here's a few to get you started:
Creekpath – Finally sold after how many millions invested
Crosswalk – Jack McDonald (aka founder of McData) SRM company that redefined the 180 degree change in direciton by becoming a clustered storage system solution who then ceased operations
Crosstor – Sold to EMC, some of the IP still lingers around
Giganet – RDMA type Ethernet based technology bought by Emulex
Highground – What did Sun ever do with it?
Imperial – One of the original SSD vendors that went throuh a few lives
Intradyn – The SMB focused email and compliance appliance reportedly recently sold at a bargin (and not to Sony)
MTI – Should have been left on Autopilot when Earl retired
Neoscale – Security and encryption assets sold to nCipher
Neustream – Open sourced based multi-protocol storage ceased operations
Sancastle – Im sure Marc Farly can tell us about that one
Sandial – IP was purchased after operatoins ceased
Stonefly – Now DNF remorphed into a storage solution vendor
Ouch, some of these are bringing back a headache, oh well, have fun and see what else comes to mind.
Cheers
gs
Stephen, forgot, what about Zetera, their website is still up, however not sure if there is still a pulse or not, maybe Toigo can chime in and update us?
Stephen, forgot, what about Zetera, their website is still up, however not sure if there is still a pulse or not, maybe Toigo can chime in and update us?
Greg, you are a crazy man!
But how many of these were COOL flops? Lots just look like flops…
Greg, you are a crazy man!
But how many of these were COOL flops? Lots just look like flops…
hey Steve – it’s been a busy few months getting ready for the announcement we did yesterday so apologies that I’m jumping in on this so late.
Actually, you might be surprised to know that HP was still shipping AutoRAID (was renamed the HP Virtual Array in its third generation product) until just a couple of years ago. It was one of those things that was ahead of its time. Customers, especially database administrator’s didn’t like the idea of not knowing where their data might be at any point in time and as you know, AutoRAID moved data between RAID5 or RAID 1/0 depending on among other things if it had been written to lately.
You might also be surprised to know that we shipped over 50,000 units of AutoRAID. But what you said about it being too small for big companies and too expensive for small ones is dead on. The tradeoffs made trying to satisfy both markets with one product were in the end bad decisions. BTW, it actually moved data at a much more granular level than a LUN. Data moved between RAID 5 or RAID 1/0 in 256KB chunks – we always tried to write data to RAID 1/0 and as the data aged and wasn’t changed, it would migrate to RAID 5. So within a single LUN, you could have data in both RAID levels.
One last HP comment – Greg asked whatever happened to HP RISS. It’s alive and well. I had a name change a couple of years ago and is called the Integrated Archive Platform (see http://h18006.www1.hp.com/products/software/im/governance_ediscovery/iap/index.html).
Thanks for the post, Calvin! I hope you can understand that I have immense respect for the products here. I even bought TWO AutoRAIDs way back in the 1990s! It was ahead of its time, offering much of what companies like Compellent tout today, but it just didn’t fit in. You’re right about the DBA mistrust – I had to fight that the whole time I had these things!
Yeah, I took no offense to what you said …. none. I find your blog a breath of fresh air in the sea of Alex/Chuck-cantankerous type blogs.
I think people would be surprised that we shipped AutoRAID for more than 10 years. And frankly, if we had focused the product on one customer (scalable/large or small/inexpensive), it might have survived the portfolio rationalization that took place when HP acquired CPQ. Then again, thinking about who the management was then (now at EMC), it had no chance of survival.
Yeah, I took no offense to what you said …. none. I find your blog a breath of fresh air in the sea of Alex/Chuck-cantankerous type blogs.
I think people would be surprised that we shipped AutoRAID for more than 10 years. And frankly, if we had focused the product on one customer (scalable/large or small/inexpensive), it might have survived the portfolio rationalization that took place when HP acquired CPQ. Then again, thinking about who the management was then (now at EMC), it had no chance of survival.
Wow, I managed to work at two of these – Mango and Revivio. Not sure if that means I’m doing something right or wrong.