Phytech Admin
asked on
VMWare Memory Issue with Heavy SQL Server queries
Scenario: We have 2 guest systems one is a file sharing server, and the other is a SQL server. Both systems, have memory allocated, and are attached to a SAN.
Problem: We are running intense queries to this SQL Server also we are importing information from files that are located on the File Server into a SQL database that is located on another ESX server.
Issue:
What we are seeing is the file that we are trying to import into our SQL database *(SQL Server is located on a Different ESX Server).
Is getting Information injected from the intense query.
NOTE: The only file with the injected information is the one that is being loaded into memory.
If we reboot the File Server we can go back and check the file in question and see that the file will no longer have the injected information.
We can also stop all queries and imports and only execute the queries. Once the memory block, for which the file that has the injected information gets over written, we can then open the file in question and see that the injected information will no longer be there.
Has anyone seen this before any help would be great?
Thanks
Problem: We are running intense queries to this SQL Server also we are importing information from files that are located on the File Server into a SQL database that is located on another ESX server.
Issue:
What we are seeing is the file that we are trying to import into our SQL database *(SQL Server is located on a Different ESX Server).
Is getting Information injected from the intense query.
NOTE: The only file with the injected information is the one that is being loaded into memory.
If we reboot the File Server we can go back and check the file in question and see that the file will no longer have the injected information.
We can also stop all queries and imports and only execute the queries. Once the memory block, for which the file that has the injected information gets over written, we can then open the file in question and see that the injected information will no longer be there.
Has anyone seen this before any help would be great?
Thanks
How have you configured your LUNs ? Is the LUN visible to both ESX Hosts ?
ASKER
Yes, LUNs are configured for Both ESX servers, the imported flat-files are being read from the LUN on the files server. We are runing iSCSI with SATA Drives
ASKER
I failed to mention that the 2 Guest OS Systems are located on the same ESX Server.
The server, where we are running the import from is on another SQL Server, importing into a different database. This SQL server is located on another ESX Server.
So that theres no confusion, file means flat-files i.e. (row of data)
Please let me know if i need to put more information.
Thanks
The server, where we are running the import from is on another SQL Server, importing into a different database. This SQL server is located on another ESX Server.
So that theres no confusion, file means flat-files i.e. (row of data)
Please let me know if i need to put more information.
Thanks
I don't know what you are talking about specifically, but have you reindexed your DBs recently? That normally sorts out speed issues (well I'm a SQL noob) but it works for our SQL servers. We don't have issues like what you mention though.
Also have you followed best practices for configuring SQL in VM?
http://communities.vmware. com/docs/D OC-8964
http://technet.microsoft.c om/en-gb/l ibrary/cc9 66534.aspx
Another thing is that I think you may be hitting an IO limit with your storage (ISCSI with SATA). SATA itself is not great for heavy loads, proven, though it works. It depends on how your underlying storage on which your VMFS volume is configured, ex how many spindles, etc.
Do you have dedicated NICs on your ESX Hosts for ISCSI traffic (with hopefully a dedicated switch for ISCSI traffic) or are all your guests and server sharing the same NIC ports, etc ... this could also be another area to look at.
Also have you followed best practices for configuring SQL in VM?
http://communities.vmware.
http://technet.microsoft.c
Another thing is that I think you may be hitting an IO limit with your storage (ISCSI with SATA). SATA itself is not great for heavy loads, proven, though it works. It depends on how your underlying storage on which your VMFS volume is configured, ex how many spindles, etc.
Do you have dedicated NICs on your ESX Hosts for ISCSI traffic (with hopefully a dedicated switch for ISCSI traffic) or are all your guests and server sharing the same NIC ports, etc ... this could also be another area to look at.
ASKER
We have dedicated NICs on the ESX hosts for ISCSI traffic and a dedicated switch,
all are guests are using one nic for the data network traffic they are not connected to the iSCSI network.
all are guests are using one nic for the data network traffic they are not connected to the iSCSI network.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.