Tweets from a NetApp Insight briefing event showed the company is preparing a replicating, deduping, clustering FlashRay iteration, complete with Glacier-targeting for StorageGRID. FlashRay, NetApp's coming ground-up designed all-flash array, will have global deduplication. Here are three tweets: #NetApp FlashRay dedupe is …
8 billion pfft
That's not even big enough for my porn collection.
I have a couple friends at DDN who recently announced(I think) something that can go to 32 trillion objects (up to 1 trillion on up to 256 servers).
"This includes object retrieval of 256 million objects per second and throughput of over 10 terabytes per second with an object retrieval latency of less than 50 milliseconds."
Now that can hold my collection.
Netapp keep telling us how bad LUN's are ?
Funny that, they're starting to sound like EMC, what with the social medial hype and constant flip flopping.
One guy's opinion in the blogosphere is not representative of an entire company's view. The good thing about Netapp is that it doesn't apply sensorship nor does it restrict its employees from having their own public opinions.
Good or bad, LUNs will be around for while and so will tape contrary to what some folks like to believe.
I didn't suggest they were attempting to censor or restrict their employees opinions, but their messaging on NFS good, LUN's (block) bad is nothing if not consistent.
Hence the slight chuckle at the irony of their messaging on the new improved all Flash Raymondo. It seems proper blocks (LUN's) are important when you really do need performance :-)
0 overhead just because its "in the OS" (in the kernel?)? Really? 0 overhead AND inline dedupe. I'd like to know how it is possible that they get the best of both worlds there! Something doesn't add up, unless the data is just not committed to disk before responding to the application, in which case it's not a viable solution for most.
Re: 0 overhead?
In Data ONTAP, writes are acknowledged to the client when the data is committed *to NVRAM*. We don't have to wait for the data to go to disk; that happens at the next Consistency Point (CP).
FlashRay runs a different O/S, but the fundamentals would be the same in this case: so long as the data is somewhere non-volatile, you can happily acknowledge to the client that it's been written.
Disclaimer: I am a NetApp employee.