I recently started doing some serious experimenting with replication jobs. Everything I have tested thus far performs as expected. However, the replication job for our production file server takes an unexpected amount of time. It takes 2 - 3 hours to replicate less than 50 Gb of changes over a gigbit circuit. The job stats show that it is using CBT, so I wouldn't expect the performance to be this slow. It also shows that the bottleneck is source, which seems to be standard as our backup jobs show similar statistics. Where it seems to hang is the 2.8 Tb data drive on the VM, 90%+ of the processing time is spent on this drive. For example, in a 3 hour 17 minute run last night 3 hours 5 minutes was spent processing that single drive (1/22/2018 6:05:55 PM :: Hard disk 2 (2.8 TB) 158.1 GB read at 15 MB/s [CBT] ----> 03:05:24). This seems a bit extreme to me.
For some background - The source volume is being read via SAN mode (1/22/2018 6:02:30 PM :: Using source proxy 192.168.250.242 for disk Hard disk 2 [san]). The volume is hosted on a Compellent SC4020 attached via 10 Gb iSCSI, so no storage integrations. However, unless I'm mistaken storage integration would only effect the snapshot process. What should I be looking at to try and improve this?
-
- Enthusiast
- Posts: 48
- Liked: 3 times
- Joined: Mar 18, 2011 7:36 pm
- Full Name: Sean Conley
-
- Expert
- Posts: 176
- Liked: 30 times
- Joined: Jul 26, 2018 8:04 pm
- Full Name: Eugene V
- Contact:
Retention policy for deleted items behavior
What kind of statistics for latency and throughput does Dell Storage Manager show for the source volume during the backup? 15MB/s does feel slow, but we may not know your baseline performance.
-
- Product Manager
- Posts: 8191
- Liked: 1322 times
- Joined: Feb 08, 2013 3:08 pm
- Full Name: Mike Resseler
- Location: Belgium
- Contact:
Re: Slow Replication Performance
Hi Sean,
I think what evilaedmin proposes is a good idea. It seems a bit slow to me also, specially when you don't have those issues with other servers. But as he said, it might be something normal or there is an issue in the backend.
As you indicated, storage integrations is for the snapshot process indeed. If you can't find out in Dell's tools if there is a problem, you can always contact our support to look deeper into these logs. Maybe they can see somewhere an issue why it is slower compared to other servers that are processed.
PS: I am assuming that other servers also have volumes on that Compellent?
I think what evilaedmin proposes is a good idea. It seems a bit slow to me also, specially when you don't have those issues with other servers. But as he said, it might be something normal or there is an issue in the backend.
As you indicated, storage integrations is for the snapshot process indeed. If you can't find out in Dell's tools if there is a problem, you can always contact our support to look deeper into these logs. Maybe they can see somewhere an issue why it is slower compared to other servers that are processed.
PS: I am assuming that other servers also have volumes on that Compellent?
Who is online
Users browsing this forum: Bing [Bot], Google [Bot], Semrush [Bot] and 54 guests