We just need the damn Google self-driving cars to get here pronto and put this discussion to bed. I'm sick of seeing idiots texting in their car, reading the paper, or whatever. I'd love to have a few pints after work and be able to go home in my own car. Provided it drove itself.
11 posts • joined 8 Jul 2013
Re: Where's the corresponding tape silo?
These systems have replicas and for serious outfits they have multiple geographies with rules in place to make sure there are always replicas in multiple geographies. If the burly men can be in several places with sledgehammers and transit vans, you are in big enough trouble that backups aren't a concern. A good lawyer or a fake passport are in order.
Re: I don't geddit.
@ratfox - or have an open system of relays which do not depend on any single person/corporation to hold the data or make the rules. We used to have this in NNTP and usenet but eventually bots, spammers and dumbasses made it too hard to find anything useful and ISPs no longer offered it is a service so you had to find a public source.
Re: out of interest...
NetApp has virtual editions of both their FAS line (7mode and clustered) and the Eseries. They are downloadable from the support site. Hitachi had a virtual VSP program that I had a copy of but I haven't looked recently. Other people pointed out the EMC emulators.
All of them are great ways to try out the interfaces for different vendors. It's also a great way to test out scripts or orchestration flows without borking your real SAN.
HDS software has nearly universally sucked forever. (I'm sure that'll register as a terrible example of grammar on any side of the Atlantic.) The 9900/USP/USPV/VSP are all rock-solid boxes that are a pain to configure but then run forever. There have been promises of better software for years and while they are better now they don't compare to the engineering found in the arrays they monitor and manage.
Do you think those CPUs in storage arrays are there just to help keep the HBAs cozy and warm? It takes compute power to run a real array and host IO directed at it and probably more so for a virtual array. There's no free lunch! You'll have to take your existing compute needs and then add the storage load onto that. You aren't going to be able to squeeze VSAN into an already fully loaded compute farm. You can try, but you'd better have your resume freshened up first or a forgiving manager with the Dell rep's number on speed dial.
Re: Dynamic provisioning
The majority of companies have similar daily/weekly io usage patterns. The cloud providers have to build based on peak usage. A bit like highways. Not everyone works 8-5 but enough do that the highways are built to accommodate that. You'll be paying for cloud providers to overbuild as well. Just like the rest of the IT world does. No free lunches.
Need another source
Big players in storage generally require dual sources for drives. Until another 2.5inch 600g 15k rpm drive comes around from another source I don't expect to see a shelf of them on many vendors' price lists.
Re: Am i the only one
Maybe it's not your age. The CRTs have all disappeared. :)
Re: what, more hardware on a monolithic processing scheme?
You are apparently misinformed. NetApp has been breaking out processes from Kahuna for years to get better use of cores. In my experience, almost every 8.x release has been a bit better in utilizing cores. Perhaps you're thinking of VNX2 which finally uses those multicore procs unlike the VNX they sold people as recently as August.
NetApp has had synchronous replication for years in 7-mode.
This sounds like it will make it easier for EMC's frenemy Cisco to bring the EMC stack further into Cloupia. Why would you want to automate the storage stack when you can bring it into the whole stack? I also heard mention of management but what of measurement and reporting? Will performance and reporting APIs for EMCs wide array of products be exposed so third parties won't need a proxy?
I get there's the file/object aspect of ViPR but that might be exposed by Openstack anyway won't it? It seems like EMC saw itself being painted into a corner with their many closed APIs and figured they'd have to open up to play in the new private cloud datacenter so they might as well productize it. Hence ViPR.
Everyone else who isn't already open with their APIs is probably thinking the same thing but EMC likes to pre-announce products far ahead of reality so they look visionary.