ChrisR
ClioSport Club Member
Which office you off to, Bedfont?
ClioSport.net
Register a free account today to become a member!
Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
-
When you purchase through links on our site, we may earn an affiliate commission. Read more here.
Upgrade Windows Server
- Thread starter Cliotoby
- Start date
Which office you off to, Bedfont?
It will be interesting to see what they say about dedupe, the v7000 doesn't have dedupe built in, it uses compression. It can use dedupe with the fastback backup product which dedupes backup data, so not dedupe on live data.
the other interesting thing will be that IBM oem netapp kit and sell it as the N series, so it's a rebadged netapp, and netapp supported dedupe and compression, on live primary data.
a
s I'm sure they'd rather you had an IBM product than a rebadged netapp one, I'd say they will say that dedupe on primary data is bad, it slows down the system and doesn't provide much benefit, not as much as compression. Which is quite frankly wrong, as with virtualisation as I mentioned you get great benefits, and then for the stuff that doesn't dedupe as well, such as databases, you can use compression, or even both if it works for the data set, and netapp supports inline and post compression. Compression does have a slight impact on performance, dedupe doesnt, it can infact speed up access, as if you have blocks that are accessed frequently, and referenced by multiple systems, those blocks can be kept in cache and so the San doesn't even need to go to disk to serve it out, cache is much much faster than accessing disk.
So yes, it will be interesting to hear what their 'take' is on it, especially with the N series being part of their portfolio.
the other interesting thing will be that IBM oem netapp kit and sell it as the N series, so it's a rebadged netapp, and netapp supported dedupe and compression, on live primary data.
a
s I'm sure they'd rather you had an IBM product than a rebadged netapp one, I'd say they will say that dedupe on primary data is bad, it slows down the system and doesn't provide much benefit, not as much as compression. Which is quite frankly wrong, as with virtualisation as I mentioned you get great benefits, and then for the stuff that doesn't dedupe as well, such as databases, you can use compression, or even both if it works for the data set, and netapp supports inline and post compression. Compression does have a slight impact on performance, dedupe doesnt, it can infact speed up access, as if you have blocks that are accessed frequently, and referenced by multiple systems, those blocks can be kept in cache and so the San doesn't even need to go to disk to serve it out, cache is much much faster than accessing disk.
So yes, it will be interesting to hear what their 'take' is on it, especially with the N series being part of their portfolio.
CSV v2 is redesigned from what was shipped with 2008 R2, I don't know the specifics of why deduplication support wasn't included but I suspect there is some interoperability issues with CSVFS.I'm not confusing anything mate, trust me, I know about this stuff, it's my life at work
I was purely talking about exchange as an example of ms removing sis from the product and as an example as to why a San supporting dedupe would be useful in that situation.
i also understand the benefits of dedupe in 2012, but I'm looking at it from a corporate pov, in an enterprise environment, and as pointed it in the link, it has some caveats, such as not being supported on csv with hyperv, that's their own product. Why doesn't it support it, performance issues? As its a cluster maybe there's an issue with who holds the fingerprint database or what ever it uses?
it just doesn't have enterprise features, it's very smb really. In your example, if y had a San with dedupe you'd gain savings on everything hosted on it, not individual servers. I'm not sure what a repro machine is or what you exactly mean by having parent disks stored on it.
like I said though, dedupe has been in windows storage server for years, this is just a new iteration of it, but then not many people use physical file servers these days, it's either hosted virtually (or being moved to that platform) or you use a San (like netapp or emc) to serve the files directly from that, a netapp San will even join the domain as a member server.
i don't think it will be long before Hp have dedupe in their sans tbh, they have it in their virtual libraries for backup but have always maintained that you don't need it in primary storage, but these days it's high on people's priority lists when looking at San features, so I think it will be a priority to get it working on their sans before too long.
With regards to repro machine I am referring to a machine I have here used to reproduce customer issues. In order to quickly get additional machines built into Hyper-V I first build the reference O/S + service pack, sysprep it and mark it read only. That is then the parent disk for any new machine I need to build using Hyper-V differencing disks. As these take up a lot of duplicate space (I have 2000 through to 2012, all SKUs) I simply turn on deduplication. I don't need a SAN, this is my local machine I need for my troubleshooting.
Similar threads
