AIM Initialization Time Estimates
Latency affects performance. How far AIM destination volumes fall out of synchronization is determined by several factors, such as the amount of I/O writes, throughput, network QOS policies, replication schedules or throttles, and network outages. Sufficient throughput (bandwidth) usually presents the biggest challenge in asynchronous IP mirroring.
Best Practice Recommendations for Initialization
-
If multiple volumes are involved then phase the initial synchronization of source and destination virtual volumes.
-
Properly configure AIM buffers:
-
-
Do not put the buffer on the boot disk.
-
Configure the buffer on standalone volumes in partitions.
-
Configure the buffer volumes to be as large as the size of the largest volume to be initialized and large enough to absorb data changes due to LAN/WAN latency delays, disconnects, etc.
-
Refer to AIM Buffers for more information.
-
Establishing Throughput Requirements
To best gauge a satisfactory throughput, you will need to quantify the change rate or the cumulative amount of data writes that occur over a period of time for the volumes you need to actively replicate. It is determined by such factors as production or peak hours and your target Recovery Point Objective ( RPO, the rollback point, or acceptable lag time).
For instance, if your organization is 24/7, has a very aggressive RPO and you’re changing a lot of data on an hourly basis, you’ll need more throughput. On the other hand, if most of your data writes occur between 8am and 5pm, and your RPO is 24 hours, and you only have about 20 GB of data changing daily, then you can tolerate a 24 hour replication latency window. You then determine you only need about a 3 MB/s connection between the two sites and the replication will “catch up” during the off-hours.
Time Estimates
If you have limited throughput and very large volumes of data to replicate, the initialization step required to synchronize the source and destination virtual volumes may take a long time to complete. The throughput of channels used for I/O writes must be sufficient; otherwise, the responsiveness of your systems will degrade.
The following table provides approximate estimates of how long it would take to initialize an AIM set using various network technologies. These estimates are based on initializing one AIM set at a time and presuming that each has dedicated use of the throughput*. Although not usually a real life scenario, this table should give you an idea of the time needed to synchronize your source and destination data over the network.
Estimated Number of Hours Needed to Replicate Data | ||||||
Technology |
20 GB |
80 GB |
120 GB |
200 GB |
300 GB |
730 GB |
T1 |
42.33 |
169.31 |
253.97 |
423.28 |
634.92 |
1544.97 |
10Base-T LAN |
6.50 |
26.01 |
39.01 |
65.02 |
97.52 |
237.31 |
DS3 (T3) |
1.50 |
6.02 |
9.03 |
15.05 |
22.57 |
54.93 |
100Base-T LAN |
0.65 |
2.60 |
3.90 |
6.50 |
9.75 |
23.73 |
OC3 |
0.42 |
1.68 |
2.52 |
4.19 |
6.29 |
15.31 |
OC12 |
0.10 |
0.42 |
0.63 |
1.05 |
1.57 |
3.82 |
*Throughput speed was calculated at 70% of full theoretical throughput allowing for application, system and network overhead factors that prohibit the full theoretical throughput of the connection from being achieved. It is not unusual for this overhead to cost 30% of your usable throughput.