Changes between Version 19 and Version 20 of LLM
- Timestamp:
- 10/13/11 16:23:47 (13 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
LLM
v19 v20 27 27 Similar to the migration of CPU/memory/disk updates, the migration of service requests is also done in an asynchronous manner, i.e., the primary machine resumes its service without waiting for an acknowledgement from the backup machine. 28 28 29 [[Image(figure2.jpg)]] 30 31 Figure 2. Checkpointing Sequence. 32 29 {{{ 30 #!html 31 <center> 32 <img style="border:0px;" src="/trac/fgbi/raw-attachment/wiki/LLM/figure2.jpg"> 33 <h4>Figure 2. Checkpointing Sequence.</h4> 34 </center> 35 }}} 36 33 37 Figure 2 shows the time sequence of migrating the checkpointed resources and the incoming service requests at different frequencies on a single network socket. The entire sequence within an epoch is described as follows: 34 38 … … 49 53 50 54 == Evaluation Results == 51 [[Image(figure3.jpg)]] 52 53 Figure 3. Downtime under highnet and highsys. 54 55 {{{ 56 #!html 57 <center> 58 <img style="border:0px;" src="/trac/fgbi/raw-attachment/wiki/LLM/figure3.jpg"> 59 <h4>Figure 3. Downtime under highnet and highsys.</h4> 60 </center> 61 }}} 62 55 63 Figure 3 shows the downtime results under highnet and highsys. We observe that under highsys, [wiki:LLM LLM] incurs a downtime that is longer than, yet comparable to, that of [http://nss.cs.ubc.ca/remus/ Remus]. The reason is that, [wiki:LLM LLM] runs at low frequency, hence the migration traffic in each period is higher than that of [http://nss.cs.ubc.ca/remus/ Remus]. Under highnet, the downtime of [wiki:LLM LLM] and [http://nss.cs.ubc.ca/remus/ Remus] show a reverse relationship, where [wiki:LLM LLM] outperforms [http://nss.cs.ubc.ca/remus/ Remus]. This is because, from the client side, there are too many duplicated packets to be served again by the backup machine in [http://nss.cs.ubc.ca/remus/ Remus]. In [wiki:LLM LLM], on the contrary, the primary machine migrates the requested packets as well as boundaries to the backup machine, i.e., only those packets yet to be served are served by the backup. Thus the client does not need to re-transmit the requests, and therefore experiences a shorter downtime. 56 64 57 65 [[Image(figure4.jpg)]] 58 59 Figure 4. Network Delay under highnet and highsys. 66 {{{ 67 #!html 68 <center> 69 <img style="border:0px;" src="/trac/fgbi/raw-attachment/wiki/LLM/figure4.jpg"> 70 <h4>Figure 4. Network Delay under highnet and highsys.</h4> 71 </center> 72 }}} 60 73 61 74 Figure 4 shows the network delay results under highnet and highsys. In both cases, we observe that [wiki:LLM LLM] significantly reduces the network delay by removing the egress queue management and releasing responses immediately. In Figure 4, we only recorded the average network delay in a migration period. Next, we show the details of the network delay in a specific migration period in Figure 4, in which the interval between two adjacent peak values represents one migration period. We observe that the network delay of [http://nss.cs.ubc.ca/remus/ Remus] decreases linearly within a period but remains at a plateau. In [wiki:LLM LLM], on the contrary, the network delay is very high at the beginning of a period, then quickly decreases to nearly zero after a system update is over. Therefore, most of the time, [wiki:LLM LLM] demonstrates a much shorter network delay than [http://nss.cs.ubc.ca/remus/ Remus]. 62 75 63 [[Image(figure5.jpg)]] 64 65 Figure 5. Overhead under Kernel Compilation. 76 {{{ 77 #!html 78 <center> 79 <img style="border:0px;" src="/trac/fgbi/raw-attachment/wiki/LLM/figure5.jpg"> 80 <h4>Figure 5. Overhead under Kernel Compilation.</h4> 81 </center> 82 }}} 66 83 67 84 Figure 5 shows the overhead under kernel compilation. The overhead significantly changes only in the checkpointing period interval of [1;60] seconds, as shown in the figure. For checkpointing in the shorter periods, the migration of system updates may last longer than a configured checkpointing period. Therefore the kernel compilation time for these cases are almost the same with minor fluctuation. For checkpointing in the longer periods, especially when it is longer than the baseline (i.e., kernel compilation without any checkpointing), a VM suspension may or may not occur during one compilation process. Therefore, the kernel compilation time will be very close to the baseline, meaning a zero percent overhead. Right in this interval, [wiki:LLM LLM]’s overhead due to the suspension of domain U is significantly lower than that of [http://nss.cs.ubc.ca/remus/ Remus], as it runs at much lower frequency than [http://nss.cs.ubc.ca/remus/ Remus].
