February 8, 2006 at 4:02 pm
February 8, 2006 at 9:53 pm
Hi Krissy,
Here's my wild stab at it. I left in my notes as I walked through it. Please let me know if I'm glaringly off: CONSTRUCTIVE criticism is always best 🙂
Network bandwidth is usually measured in bits (e.g., Mb or Gb) so, let's convert our Gigabyte (GB) data units accordingly (all the way to Mega so we can compare against our available bandwidth, also in Mega-units)...
1.5GB * 8 = 12Gb = 12000Mb (amount of data we need to transfer in 15 min.)
12000Mb / (15 * 60) = 13.3 Mb/S (divide our file size by (15 minutes * 60 to get seconds) to get our bandwidth requirement for a 15 minute transfer)
13.3 * 1.2 = 15.96 (13.3Mb/S is our raw bandwidth requirement. Let's assume 80% bandwidth utilization since you rarely get 100% [other traffic, packet collisions, etc])
you would need 16Mb/S bandwidth to transfer 1.5GB within a 15 minute window.
If your connection is really 4MB/S (32Mb) then you should have no problem.
However, if it's 4Mb/S (I only throw this out as a possibility because MB is used so often in place of Mb) then you won't make it without compressing your backup files (I typically get about 75% compression of sql backup files using gzip. This could get you very close to a 4Mb/S bandwidth requirement for just the data transfer. But, you'd still miss your window due to the amount of time it takes to compress the file. If you use a backup compression tool (RedGate Backup comes right to mind; there are others), you may gain enough time in compression to still have a chance of making your transfer window.
February 9, 2006 at 2:43 pm
February 9, 2006 at 3:31 pm
February 9, 2006 at 7:22 pm
Yes, the /S is per second.
Your calculations look good but... I'd guess that your file is probably sized using GB. Just to confuse people, storage is typically measured in Byte units while networking it meaured in Bit units. You can right click and view properties on your file to make sure. If you're off, it's going to make a big difference!
Just so you don't have to worry about mistakes determining what would work "theoretically," you could always copy your tran log from production to DR and time it (assuming you have your file and link available for testing)
February 9, 2006 at 8:42 pm
February 9, 2006 at 10:40 pm
Hi Ken,
I'm interested on the formula you used (wanted to learn how)
why did you use the factor 8 to get total GB and 1.2 to get raw bandwidth rate?
thanks in advance
February 10, 2006 at 3:41 am
Hi Krissy,
depending on the way you implement the logshipping you might want to time the restore into the db too. If copying and restoring is done from server1 you have to add the transfer time and the restore time...
karl
Best regards
karl
February 10, 2006 at 6:17 am
Are you saying that your transaction log is currently 1.5GB, or that 1.5GB of log is produced every 15 minutes? If it's the former, once the initial transfer has been done you shouldn't have a problem as the incremental files wil be a lot smaller.
--
Scott
February 12, 2006 at 3:25 pm
I know I need to take this into account, but if it was going to take longer than 15 minutes just to copy the logs over then I obviously won't bother with that frequency. Just wasn't sure how to work it out!
February 12, 2006 at 3:42 pm
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply