Changing heap size parameters
By default, DataStax Enteprise (DSE) sets the Java Virtual Machine (JVM) heap size from 1 to 32 GB depending on the amount of RAM and type of Java installed.
The cassandra-env.sh
automatically configures the min and max size to the same value using the following formula:
max(min(1/2 ram, 1024 megabytes), min(1/4 ram, 32765 megabytes))
To adjust the JVM heap size, uncomment and set the following parameters in the jvm-server.options
file:
-
Minimum (
-Xms
) -
Maximum (
-Xmx
) -
New generation (
-Xmn
) -
Parallel processing for GC (
-XX:+UseParNewGC
) located injvm8-server.options
orjvm11-server
.options depending upon the JDK.
Restriction: When overriding the default setting, both min and max must be defined the jvm-server.options
file.
Additionally, for larger machines, increase the max direct memory (-XX:MaxDirectMemorySize
), but leave around 15-20% of memory for the OS and other in-memory structures.
Guidelines and recommendations
Setting the Java heap higher than 32 GB may interfere with the OS page cache. Operating systems that maintain the OS page cache for frequently accessed data are very good at keeping this data in memory. Properly tuning the OS page cache usually results in better performance than increasing the row cache. For production use, follow these guidelines to adjust heap size for your environment:
-
Heap size is usually between ¼ and ½ of system memory but not larger than 32 GB.
-
Reserve enough memory for the offheap cache and file system cache.
-
Enable GC logging when adjusting GC.
-
Gradually increase or decrease the parameters. Test each incremental change.
-
Enable parallel processing for GC, particularly when using DSE Search.
-
The GCInspector class logs information about any garbage collection that takes longer than 200 ms. Garbage collections that occur frequently and take a moderate length of time (seconds) to complete indicate excessive garbage collection pressure on the JVM. In addition to adjusting the garbage collection options, other remedies include adding nodes, and lowering cache sizes.
-
For a node using G1, DataStax recommends a
MAX_HEAP_SIZE
as large as possible, up to 64 GB.
For more tuning tips, see Secret HotSpot option improving GC pauses on large heaps. |
Maximum and minimum heap size
The recommended maximum heap size depends on which GC is used:
Hardware setup | Recommended MAX_HEAP_SIZE |
---|---|
G1 for newer computers (8+ cores) with up to 256 GB RAM |
16 GB to 32 GBSee Java performance tuning. |
CMS for newer computers (8+ cores) with up to 256 GB RAM |
No more than 16 GB |
Older computers |
Typically 8 GB |
New heap size
For CMS, you may also need to adjust new (young) generation heap size. This setting determines the amount of heap memory allocated to newer objects. The database calculates the default value for this property in megabytes (MB) as the lesser of:
-
100 times the number of cores
-
¼ of MAX_HEAP_SIZE
Procedure
-
To enable GC logging, uncomment the
loggc
parameter located injvm8-server.options
orjvm11-server.options
depending upon the JDK.-Xloggc:/var/log/cassandra/gc.log
After restarting Cassandra the log is created and GC events are recorded.
-
Set the heap sizes in the
jvm-server.options
file:-
Uncomment and set both the min and max heap size. For example to set both the min and max heap size to 16 GB:
-Xms16G -Xmx16G
Set the min (-Xms) and max (-Xmx) heap sizes to the same value to avoid stop-the-world GC pauses during resize, and to lock the heap in memory on startup which prevents any of it from being swapped out.
-
If using CMS, uncomment and set the new generation heap size to tune the heap for CMS. As a starting point, set the new parameter to 100 MB per physical CPU core. For example, for a modern eight-core or greater system:
-Xmn800M
A larger size leads to longer GC pause times. For a smaller new size, GC pauses are shorter but usually more expensive.
-
-
On larger machines, increase the max direct memory (
-XX:MaxDirectMemorySize
), but leave around 15-20% of memory for the OS and other in-memory structures. For example, to set the max direct memory to 1 MB:-XX:MaxDirectMemorySize=1M
By default, the size is zero, so the JVM selects the size of the NIO direct-buffer allocations automatically.
Alternatively, you can set an environment variable called
MAX_DIRECT_MEM
, instead of setting a size for-XX:MaxDirectMemorySize
in thejvm-server.options
file. -
Save and close the
jvm-server.options
file. -
Restart Cassandra and run some read heavy or write heavy operations.
-
Check the GC logs.
This method decreases performance for the test node, but generally does not significantly reduce cluster performance.
If performance does not improve, contact the DataStax Services team for additional help.