![]() |
![]() HP OpenVMS Systemsask the wizard |
![]() |
The Question is: Volume shadowing in a multiple site cluster. Are there any guidelines to find out how big the network connection needs to be. We upgrading an existing VAX cluster to Alpha and are going to add some DR functions to it at the same time. I'm trying to find the best solution to keep the data syncronized between the sites and the cost for the options. I have looked for some kind of s ize info or guidelines for the size of the network pipe between sites but can't find anything beyond "at least 10MBit". The Answer is : You need to determine how much bandwidth you need and particularly how quickly you need to resynchronize the data assuming host-based volume shadowing and particularly volume merges or volume copies The traditional minimum for clustering bandwidth and for latency is that of an IEEE 802.3 or Ethernet network; 10 megabits per second. This is the lower network performance bound for a supported cluster configuration. For remote (or disaster-tolerant; DT) connections, this implies T3 or E3 connection speeds, or faster. How fast a connection is required for your configuration depends on how many volumes are involved and how fast you need to (re)synchronize the contents of the shadowset volumes. This is in addition to the steady-state network I/O performance requirements, of course. DT configuration design might initially looks comparatively trivial, but is in actuality surprisingly difficult to complete correctly and reliably -- the dependencies and the subtleties are many and varied. HP services has extensive experience in these and offers DT Services, and the OpenVMS Wizard strongly encourages any customer seeking to create a first-time DT configuration avail themselves of these HP Services DT Services offerings.
|