Replicating across the void
Our DR EqualLogic has relocated to our remote site and is up and running. there is a whole bunch of networking that need to be done, but now the we have Dell Sumo unit on the end of a 1Gbps link and our production units at the other end with replication going between the two.
Somehow between BT and Virgin Media we have managed to get jumbo frames working between the two sites and with 3ms delay each way on the 1Gig link we actually saw it hitting the full 1Gbps speed, which is not what we would have expected, anything in the 5-800 would have been desirable, but with the 6ms round trip we were expecting to see around 250 or less certainly our testing with WANem indicated a much slower replication, though that was with 5 EqualLogic units (1 SATA and 4 SAS) replicating around 20 volumes. Replicating just the one or two volumes saw a much lower transfer rate. We will have to see how it goes over the next few weeks, especially once the VMS boys start replicating their EVA stuff.
The initial catch up replication of around 1.5TB took about 7 hours to complete, which is about 215 GB and hour, which averages out at around 476Mbps, if my calculator finger is working correctly.
Speaking of EVAs, I had the pleasure of removing all the fibre from under the floor in the server room to day. Just the one server still connected to the EVA, that is the previously mentioned developer, who still prefers Direct attached storage then EVA, and then iSCSI. It took us about 3 years to persuade them SANs were OK, but he still feels uncomfortable, every step is a painful drag. He always manages to find a someone on the web to support his view. He is a luddite.
Somehow between BT and Virgin Media we have managed to get jumbo frames working between the two sites and with 3ms delay each way on the 1Gig link we actually saw it hitting the full 1Gbps speed, which is not what we would have expected, anything in the 5-800 would have been desirable, but with the 6ms round trip we were expecting to see around 250 or less certainly our testing with WANem indicated a much slower replication, though that was with 5 EqualLogic units (1 SATA and 4 SAS) replicating around 20 volumes. Replicating just the one or two volumes saw a much lower transfer rate. We will have to see how it goes over the next few weeks, especially once the VMS boys start replicating their EVA stuff.
The initial catch up replication of around 1.5TB took about 7 hours to complete, which is about 215 GB and hour, which averages out at around 476Mbps, if my calculator finger is working correctly.
Speaking of EVAs, I had the pleasure of removing all the fibre from under the floor in the server room to day. Just the one server still connected to the EVA, that is the previously mentioned developer, who still prefers Direct attached storage then EVA, and then iSCSI. It took us about 3 years to persuade them SANs were OK, but he still feels uncomfortable, every step is a painful drag. He always manages to find a someone on the web to support his view. He is a luddite.
Comments
Post a Comment