Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
markus876
Aug 19, 2002

I am a comedy trap.

Corvettefisher posted:

NFS - Network file share, one of the slower storage techniques but very cheap and easy to maintain, can be greatly improved by use of jumbo frames. TCP/IP based Speeds Vary on Network Speed, latency is a bit higher than others
iScsi - Block level access to data over the network, better performance than NFS by a long shot, gets a bump from jumbo frames but isn't as noticeable as NFS gains, TCP/IP based, Speeds from 10Mb/s-10Gb/s, latency is comparable to that of a normal disk if set up on a vlan, network, or subetnet made for storage. Cheap and easy to maintain, good if you have an existing ethernet network

I'm not sure that you should be generalizing these two options like this.

For NFS, I wouldn't necessarily say it is "cheaper" (or more expensive for that matter) than iSCSI - they both generally use similar switching and cabling architectures. In fact, iSCSI may involve additional costs with specialized HBAs, although many deployments are no longer bothering with HBAs, especially with ESXi-bootable SD/USB sticks that mean you can run without any local disks on your ESXi servers and the performance of software initiators improving.

You could say that for the most part NFS (and iSCSI) both tend to have lower hardware costs than traditional FC deployments since you don't have to purchase FC switches and HBAs.

I would definitely not say that iSCSI has "better performance than NFS by a long shot".

See this TR from (admittedly-possibly-biased) Netapp: http://media.netapp.com/documents/tr-3697.pdf Yes, it's from 2008, but as far as I know things are pretty similar today still.

See also http://media.netapp.com/documents/tr-3749.pdf that goes into describing NFS best practices, etc.

To summarize - I would say that NFS is typically not noticeably faster or slower (in most cases) than iSCSI for most workloads.

I believe that using NFS is far simpler from the VMware side (adding datastores, having the possibility to move or copy around files by just mounting the NFS share on another host (this is especially great for the shared "media" datastore that you can mount NFS read-only on your ESX hosts, and mount rw on a unix workstation and copy ISOs to it for booting VMs), not worrying about block sizes, and VMFS extents, being able to resize an NFS share and instantly see the new size/capacity in VMware).

The biggest factor in choosing if you want to use NFS or iSCSI should probably just be what kind of support your storage has for NFS. If you are running on a platform that provides good NFS support, I highly recommend exploring and trying it out.

Adbot
ADBOT LOVES YOU

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply