In this blog, we’ll discuss how Percona XtraDB Cluster handled a high latency network environment.
Recently I was working in an environment where Percona XtraDB Cluster was running over a 10GB network, but one of the nodes was located in a distant location and the ping time was higher than what you would typically expect.
For example, the following shows the ping time between nodes in my local cluster:
|
1 |
ping 172.16.0.1 -s 1024<br>PING 172.16.0.1 (172.16.0.1) 1024(1052) bytes of data.<br>1032 bytes from 172.16.0.1: icmp_seq=1 ttl=64 time=0.144 ms<br>1032 bytes from 172.16.0.1: icmp_seq=2 ttl=64 time=0.110 ms<br>1032 bytes from 172.16.0.1: icmp_seq=3 ttl=64 time=0.109 ms<br>1032 bytes from 172.16.0.1: icmp_seq=4 ttl=64 time=0.125 ms<br> |
Generally speaking, you can say that the ping time was 0.1ms.
Now let’s say one node has a ping time of 7ms. Percona XtraDB Cluster with the default settings does not handle this case very well. There is, however, some good news: a small configuration change can improve things dramatically – you just need to know what to change!
Let’s review this case, and for the test I’ll use following sysbench:
|
1 |
sysbench --test=tests/db/oltp.lua --oltp_tables_count=100 --oltp_table_size=1000000 --num-threads=50 --mysql-host=172.16.0.4 --mysql-user=sbtest --oltp-read-only=off --max-time=3600 --max-requests=0 --report-interval=10 --rand-type=uniform --rand-init=on run<br> |
For the first example, all the nodes have equal latency (0.1ms), and for the second example we’ll introduce a 7ms latency to one of the nodes. You can easily do this in Linux with the following command:
|
1 |
# Add 7ms delay for network packets <br>tc qdisc add dev eno2 root netem delay 7ms<br># to remove delay use:<br># tc qdisc del dev eno2 root netem<br> |
So now let’s compare both the throughput and the response time for both cases:
Or in numbers:
| Setup | Throughput (average), tps | 95% response time (average), ms |
| No latency | 2698 | 27.43 |
| 7ms latency | 739 | 571.40 |
As you can see, that is a huge difference! The variance in throughput and response time is also significant. Apparently there are two variables responsible for that:
The evs.send_window variable defines the maximum number of data packets in replication at a time. For WAN setups, the variable can be set to a considerably higher value than the default (for example, 512).
So now let’s start a cluster with --wsrep_provider_options="evs.send_window=512;evs.user_send_window=512" in the 7ms node case.
How do the throughput and response time change? Let’s see:
Or in numeric format:
| Setup | Throughput (average), tps | 95% response time(average), ms |
| No latency | 2698 | 27.43 |
| 7ms latency | 739 | 571.40 |
| 7ms latency – Optimized | 2346 | 46.85 |
We can see that there is still some performance penalty. In the end, it is impossible to mitigate the high response time from one node,
but that’s a big improvement compared to default settings.
So if you have a high-bandwidth network with a node that shows a high ping time, consider changing evs.send_window and evs.user_send_window.
Resources
RELATED POSTS
