Symantec两地三中心应用级云容灾方案汇.ppt

* * Fire Drill splits BCV snaps at remote site, mounts the BCV, then starts the SG on the mount Requires SF 5.0: BCV of a VM diskgroup will create 2 identical diskgroups; SF 5.0 renames the DG on the BCV’s so VM can import them Fire Drill SGs are isolated from production SGs, ensuring no production impact Prevents concurrency violation Walkthrough Setup: VCS HA/DR running as normal Click 1: Initiate Fire Drill. The Fire Drill Agent takes a snapshot of the replicated data (Business Continuity Volume). Click 2: Mount Snapshot. Fire Drill modifies the snapshot volume metadata (via Storage Foundation 5.0) so that the snapshot can be re-imported back on the original host. Fire Drill imports the BCV and mounts the necessary filesystems. Click 3: Test App. Fire Drill starts up the target Service Group and all its required resources. The test SG is a clone of the production SG and does not impact production. The Service Group starts successfully and we know that we are fully protected at this point in time. Click 4: Reset. Fire Drill shuts down the target SG, unmounts the snapshot and returns the snapshot storage to the array. * * * * * * * SRDF/A is a relatively new product (1.0) that has seen very little adoption. It is based on the same principles as SRDF/S, but instead of sending writes over the SAN synchronously, it holds them in the disk cache memory buffer for 30 seconds (default), then drains them over the network. SRDF/A maintains two cache buffers, so when one is draining the other is filling up. As you can imagine, problems occur if one buffer fills before another has finished draining and is ready to fill again. As a result this product is is susceptible to cache overflow and would probably be run with SRDF/DM (previously called Adaptive Copy) to protect against those times when the buffers fail. Additionally, these cache buffers will not likely be sized to handle excessive filling since they represent the most expensive resource in the storage environme