What's interesting about Hedvig?
Transcript
hi Keith Towser from the CTO at visor calm it's been awhile since I've done a coffee chat so I thought I'd do one based off of a storage field day 10 presenter which was hit HED Vig there are another software-defined storage should play I know you're saying another even at the storage field they have watched a couple of videos one was primary data there a data virtualization play I've talked about them in the past I probably do another coffee chat just to update you on their
technologies then outside of storage field day you look at the larger market you have em where visa and MC scale and eggs anta spring path maxtor the tradition of torture the traditional x86 hyper converged players and new tonics pivot three simplicity the list goes on and on and on what's interesting about heaven a couple of things right out the gate a couple of design decisions that I thought was interesting one they use proxy clients or this concept of a proxy that provides basically cash host to
the recap cash close to the application there a data set that application can be either your hypervisor so you can use this as an appliance and you have a VSA based off of heavy edged storage technology and that VSA is on a distributed set of esxi or hyper-v nose is pretty much hypervisor agnostic because it's basically a virtual machine that proxy can run on each individual esxi or hyper-v node and provide re level access or re cash to your VMS on that local storage box no
so pre-stored standard but they go a little bit further and use the proxenos to do a little bit more than that one the proxenos are aware of state of the underlying storage architecture there you have this concept called a I think it's called page system so what it allows them to do is add advanced storage features such as deduplication and compression with a little bit less cost the proxy systems since there are aware of the underlying underlay they can pre kind of call the paging system
and be aware when you go to do a write if that hash for that right already exists so you can d do basically in line will follow adding an awful lot of a latency or overhead to a system so if you want to use this in a hyper converged solution versus a hyperscale solution the hit you take for enabling something like hyper converged or I'm sorry for enabling something like inline deduplication is not as big as it would be in other solutions other things of interest
off the top I thought it was interesting that just like VMware vCenter using a proprietary technology from a storage communication layer to communicate know to know to do know to know replication uh or no to proxy storage communication if you're accessing storage or if you're accessing storage via a client directly and bypassing the pot proxy that's tradition of storage technologies but when they're doing internal communications they're using a proprietary IP layer which I thought is interesting other solutions are doing similar or taking similar approaches there
must be inherent lack of ability and something like a I scuzzy or NFS or what other what are whatever other transport is available for transferring data over a data network for storage replication so that's the high-level things that I thought with the wide range of solutions on the market the things that I thought were off the top interesting swing by see the storage filled day videos tech field day videos pretty good storage medium looks like they got new cameras because the quality ray looked really good
in it talk to you guys next tech talk follow me on the twitters at sitio advisor and the website is the CTO advisor calm
