Sorry for the newbie questions, but I haven't found direct answers in the literature or forums so thought I'd try asking here.
!. How are CPU resources used from a threading standpoint? For example, how does a small number of fast CPUs compare to a large number of slower CPUs?
2. How is RAM used? In particular, does main memory serve as a cache and what other things consume memory? If main memory acts as cache, how does this interact with cache RAM on the RAID controller card? Also, is there any guideline for a typical ratio of RAM size to disk dize?
3. What are the performance issues associated with snapshots? I see no specific limit to the number of snapshots, but what is practical? Would it be reasonable, for example, to make daily snapshots so that there are essentially 365 snapshots at all times? And how does Open-E's snapshot technology compare to something like NetApp's StoreVault which claims no performance penalty?
4. Is there any interaction with the snapshot capability of the hardware RAID controller, or are snapshots always handled internally to Open-E?
5. Are there any performance comparisons available to any other commercial hardware solutions?
6. When synchronizing across multiple systems, are changes buffered in case connectivity is lost? For example, if the network connection between the two systems is lost for several hours, once connectivity is restored, will all pending changes still be made?
I think that's about it for now. This looks like a really great product and we look forward to testing it.
1. How are CPU resources used from a threading standpoint? For example,
how does a small number of fast CPUs compare to a large number of slower
CPUs?
>> It depends on how many users are going to connect. So if there are 1-20 users then
quad core CPU will not be efficient. But if there will be more than 100 users
connected then quad core CPU will be somehow mandatory.
2. How is RAM used? In particular, does main memory serve as a cache and
what other things consume memory? If main memory acts as cache, how does
this interact with cache RAM on the RAID controller card? Also, is there
any guideline for a typical ratio of RAM size to disk dize?
>> Our software use as much available RAM for caching. The required size of the RAM
depends on the use, if you are planning to use backups, snapshots, replication etc
then you will need 4GB-8GB of RAM.
RAID controllers have their own cache, for instance 3Ware controllers have the
function to chose between security and performance. So, controllers use their
own cache.
And there is ratio between RAM and disk sizes, again it depends on the usage.
3. What are the performance issues associated with snapshots? I see no
specific limit to the number of snapshots, but what is practical? Would it
be reasonable, for example, to make daily snapshots so that there are
essentially 365 snapshots at all times? And how does Open-E's snapshot
technology compare to something like NetApp's StoreVault which claims no
performance penalty?
>> Our DSS is limited to 10 snapshot for each logical volume so this configuration is
not possible. Also, large number of snapshot will generate very high load which will
slow the performance. Open-e recommend 10-12% of the volume size for a snapshots, or
3 times the amount of changes that will be generated.
There are feature deference’s between open-s DSS and NetApp’s StorVault like
max capacity, FTP, NTP JumboFrame and more. I strongly recommend contacting
NetApp and request a demo, then compare features and costs between the two
software’s. I guarantee you that you will find that our product is more
reliable over competitors products, plus you will love the cost.
4. Is there any interaction with the snapshot capability of the hardware
RAID controller, or are snapshots always handled internally to Open-E?
>> There is no interaction between snapshots and hardware as snapshots are handled
internally.
5. Are there any performance comparisons available to any other commercial
hardware solutions?
>> We do not sell hardware’s, we just sell and market our software, but you can check
our hardware compatibility list from http://www.open-e.com/data_storage_s...ang=en&subserv
=prodcomp
And you can compare it to other solution by testing our Demo or our LITE
version (mentioned in the beginning of the e-mail)
6. When synchronizing across multiple systems, are changes buffered in
case connectivity is lost? For example, if the network connection between
the two systems is lost for several hours, once connectivity is restored,
will all pending changes still be made?
>>This one has two parts: For the iSCSI: if connection get lost, iSCSI will resynch
and continue the backup.
For the NAS part, if connection get lost NAS will kicks off in the next task
schedule time.
7. Do you have a favorite RAID controller? Our thought is to create a
single RAID 6 array of about 6-8 TB, which should be sufficient for our
needs. About half of the space will be iSCSI drives for VMware servers,
and the other half will be NAS storage.
>> You can check the compatibility of controllers from the previous link.
Has DSS V6 changed as far as usage of multiple cores is concerned?
For instance, we have three XenServers that connect to our SAN. I am looking at upgrading from ISICI-R3 to DSS V6, and want to know if it will make efficient usage of the cpu's.
There are 22 VM's (Six servers and sixteen desktops) running on the three XenServers that will connect to the SAN by 10Gb ethernet. In addition there are about twelve desktops that are not virtualized that will connect to the VM Servers.
You speak of users, but I'm not sure how that translates for us. The three servers are the only machines that connect to the SAN, and feed the data to the VM's.
The reason I ask is I am looking to find out if a Dell 2950III with Dual Quad core Xeons and 16GB Ram is overkill. If so, I'll look for something that's a bit cheaper.
resource usage & snapshot performance
Reply With Quote
