Skip to content

Commit

Permalink
Update README and multi_bench
Browse files Browse the repository at this point in the history
  • Loading branch information
@mranney
mranney committed Jul 10, 2015
1 parent a2ebe4f commit eb03bb9
Show file tree
Hide file tree
Showing 2 changed files with 148 additions and 100 deletions.
150 changes: 106 additions & 44 deletions README.md
View file
@@ -1,24 +1,14 @@
redis - a node.js redis client
===========================

This is a complete Redis client for node.js. It supports all Redis commands, including many recently added commands like EVAL from
experimental Redis server branches.

This is a complete Redis client for node.js. It supports all Redis commands,
including many recently added commands like EVAL from experimental Redis server
branches.

Install with:

npm install redis

Pieter Noordhuis has provided a binding to the official `hiredis` C library, which is non-blocking and fast. To use `hiredis`, do:

npm install hiredis redis

If `hiredis` is installed, `node_redis` will use it by default. Otherwise, a pure JavaScript parser will be used.

If you use `hiredis`, be sure to rebuild it whenever you upgrade your version of node. There are mysterious failures that can
happen between node and native code modules after a node upgrade.


## Usage

Simple example, included as `examples/simple.js`:
Expand Down Expand Up @@ -57,34 +47,9 @@ This will display:
1: hashtest 2
mjr:~/work/node_redis (master)$


## Performance

Here are typical results of `multi_bench.js` which is similar to `redis-benchmark` from the Redis distribution.
It uses 50 concurrent connections with no pipelining.

JavaScript parser:

PING: 20000 ops 42283.30 ops/sec 0/5/1.182
SET: 20000 ops 32948.93 ops/sec 1/7/1.515
GET: 20000 ops 28694.40 ops/sec 0/9/1.740
INCR: 20000 ops 39370.08 ops/sec 0/8/1.269
LPUSH: 20000 ops 36429.87 ops/sec 0/8/1.370
LRANGE (10 elements): 20000 ops 9891.20 ops/sec 1/9/5.048
LRANGE (100 elements): 20000 ops 1384.56 ops/sec 10/91/36.072

hiredis parser:

PING: 20000 ops 46189.38 ops/sec 1/4/1.082
SET: 20000 ops 41237.11 ops/sec 0/6/1.210
GET: 20000 ops 39682.54 ops/sec 1/7/1.257
INCR: 20000 ops 40080.16 ops/sec 0/8/1.242
LPUSH: 20000 ops 41152.26 ops/sec 0/3/1.212
LRANGE (10 elements): 20000 ops 36563.07 ops/sec 1/8/1.363
LRANGE (100 elements): 20000 ops 21834.06 ops/sec 0/9/2.287

The performance of `node_redis` improves dramatically with pipelining, which happens automatically in most normal programs.

Note that the API is entire asynchronous. To get data back from the server,
you'll need to use a callback. The return value from most of the API is a
backpressure indicator.

### Sending Commands

Expand Down Expand Up @@ -658,6 +623,105 @@ client.zadd(args, function (err, response) {
});
```

## Performance

Much effort has been spent to make `node_redis` as fast as possible for common
operations. As pipelining happens naturally from shared connections, overall
efficiency goes up.

Here are typical results of `multi_bench.js` which is similar to
`redis-benchmark` from the Redis distribution. It uses 5 concurrent connections
and shows the difference between pipelines of 1 and 50.

JavaScript parser:

Client count: 5, node version: 0.10.32, server version: 2.8.18, parser: javascript
PING, 1/5 min/max/avg/p95: 0/ 3/ 0.05/ 1.00 1103ms total, 18132.37 ops/sec
PING, 50/5 min/max/avg/p95: 0/ 4/ 0.81/ 2.00 327ms total, 61162.08 ops/sec
SET 4B str, 1/5 min/max/avg/p95: 0/ 2/ 0.05/ 0.00 1104ms total, 18115.94 ops/sec
SET 4B str, 50/5 min/max/avg/p95: 0/ 3/ 0.83/ 2.00 333ms total, 60060.06 ops/sec
SET 4B buf, 1/5 min/max/avg/p95: 0/ 2/ 0.09/ 1.00 1876ms total, 10660.98 ops/sec
SET 4B buf, 50/5 min/max/avg/p95: 0/ 11/ 2.55/ 4.00 1025ms total, 19512.20 ops/sec
GET 4B str, 1/5 min/max/avg/p95: 0/ 1/ 0.05/ 1.00 1117ms total, 17905.10 ops/sec
GET 4B str, 50/5 min/max/avg/p95: 0/ 3/ 0.87/ 2.00 347ms total, 57636.89 ops/sec
GET 4B buf, 1/5 min/max/avg/p95: 0/ 1/ 0.05/ 1.00 1110ms total, 18018.02 ops/sec
GET 4B buf, 50/5 min/max/avg/p95: 0/ 2/ 0.85/ 2.00 342ms total, 58479.53 ops/sec
SET 4KiB str, 1/5 min/max/avg/p95: 0/ 1/ 0.05/ 1.00 1119ms total, 17873.10 ops/sec
SET 4KiB str, 50/5 min/max/avg/p95: 0/ 3/ 0.89/ 2.00 358ms total, 55865.92 ops/sec
SET 4KiB buf, 1/5 min/max/avg/p95: 0/ 1/ 0.09/ 1.00 1894ms total, 10559.66 ops/sec
SET 4KiB buf, 50/5 min/max/avg/p95: 0/ 7/ 2.57/ 4.00 1031ms total, 19398.64 ops/sec
GET 4KiB str, 1/5 min/max/avg/p95: 0/ 6/ 0.06/ 1.00 1248ms total, 16025.64 ops/sec
GET 4KiB str, 50/5 min/max/avg/p95: 0/ 3/ 1.03/ 2.00 415ms total, 48192.77 ops/sec
GET 4KiB buf, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 1.00 1177ms total, 16992.35 ops/sec
GET 4KiB buf, 50/5 min/max/avg/p95: 0/ 10/ 1.02/ 2.00 409ms total, 48899.76 ops/sec
INCR, 1/5 min/max/avg/p95: 0/ 2/ 0.05/ 0.55 1137ms total, 17590.15 ops/sec
INCR, 50/5 min/max/avg/p95: 0/ 2/ 0.85/ 2.00 343ms total, 58309.04 ops/sec
LPUSH, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 1.00 1143ms total, 17497.81 ops/sec
LPUSH, 50/5 min/max/avg/p95: 0/ 3/ 0.87/ 2.00 350ms total, 57142.86 ops/sec
LRANGE 10, 1/5 min/max/avg/p95: 0/ 2/ 0.06/ 1.00 1283ms total, 15588.46 ops/sec
LRANGE 10, 50/5 min/max/avg/p95: 0/ 3/ 1.12/ 2.00 449ms total, 44543.43 ops/sec
LRANGE 100, 1/5 min/max/avg/p95: 0/ 1/ 0.09/ 1.00 1932ms total, 10351.97 ops/sec
LRANGE 100, 50/5 min/max/avg/p95: 0/ 5/ 2.46/ 4.00 985ms total, 20304.57 ops/sec
SET 4MiB buf, 1/5 min/max/avg/p95: 1/ 4/ 1.37/ 2.00 691ms total, 723.59 ops/sec
SET 4MiB buf, 50/5 min/max/avg/p95: 3/ 166/ 57.66/ 116.00 601ms total, 831.95 ops/sec
GET 4MiB str, 1/5 min/max/avg/p95: 84/ 110/ 93.18/ 106.95 9320ms total, 10.73 ops/sec
GET 4MiB str, 50/5 min/max/avg/p95: 156/7375/3400.10/6840.40 8928ms total, 11.20 ops/sec
GET 4MiB buf, 1/5 min/max/avg/p95: 84/ 105/ 91.21/ 99.00 9129ms total, 10.95 ops/sec
GET 4MiB buf, 50/5 min/max/avg/p95: 424/5704/3518.94/5626.65 9145ms total, 10.93 ops/sec

If you use very large responses in your application, the JavaScript parser
performs badly. Until the JS parser is fixed, you can use the C-based `hiredis`
parser bound to the official `hiredis` C library. To use `hiredis`, do:

npm install hiredis redis

If the `hiredis` npm module is installed, `node_redis` will use it by default.
Otherwise, the pure JavaScript parser will be used.

If you use `hiredis`, be sure to rebuild it whenever you upgrade your version of
node. There are mysterious failures that can happen between node and native
code modules after a node upgrade.

Most users find that the JS parser is faster than the `hiredis` parser. Because
of the pain associated with upgrading native code modules, you should only use
`hiredis` if your application needs it.

hiredis parser:

Client count: 5, node version: 0.10.32, server version: 2.8.18, parser: hiredis
PING, 1/5 min/max/avg/p95: 0/ 3/ 0.05/ 1.00 1092ms total, 18315.02 ops/sec
PING, 50/5 min/max/avg/p95: 0/ 5/ 0.87/ 2.00 347ms total, 57636.89 ops/sec
SET 4B str, 1/5 min/max/avg/p95: 0/ 2/ 0.06/ 1.00 1151ms total, 17376.19 ops/sec
SET 4B str, 50/5 min/max/avg/p95: 0/ 3/ 0.83/ 2.00 334ms total, 59880.24 ops/sec
SET 4B buf, 1/5 min/max/avg/p95: 0/ 3/ 0.09/ 1.00 1932ms total, 10351.97 ops/sec
SET 4B buf, 50/5 min/max/avg/p95: 0/ 9/ 2.64/ 4.00 1059ms total, 18885.74 ops/sec
GET 4B str, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 0.00 1185ms total, 16877.64 ops/sec
GET 4B str, 50/5 min/max/avg/p95: 0/ 3/ 0.85/ 2.00 341ms total, 58651.03 ops/sec
GET 4B buf, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 0.00 1179ms total, 16963.53 ops/sec
GET 4B buf, 50/5 min/max/avg/p95: 0/ 3/ 0.85/ 2.00 340ms total, 58823.53 ops/sec
SET 4KiB str, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 1.00 1210ms total, 16528.93 ops/sec
SET 4KiB str, 50/5 min/max/avg/p95: 0/ 3/ 0.93/ 2.00 372ms total, 53763.44 ops/sec
SET 4KiB buf, 1/5 min/max/avg/p95: 0/ 1/ 0.10/ 1.00 1967ms total, 10167.77 ops/sec
SET 4KiB buf, 50/5 min/max/avg/p95: 0/ 6/ 2.63/ 4.00 1053ms total, 18993.35 ops/sec
GET 4KiB str, 1/5 min/max/avg/p95: 0/ 6/ 0.06/ 1.00 1176ms total, 17006.80 ops/sec
GET 4KiB str, 50/5 min/max/avg/p95: 0/ 4/ 1.00/ 2.00 399ms total, 50125.31 ops/sec
GET 4KiB buf, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 1.00 1158ms total, 17271.16 ops/sec
GET 4KiB buf, 50/5 min/max/avg/p95: 0/ 3/ 0.99/ 2.00 398ms total, 50251.26 ops/sec
INCR, 1/5 min/max/avg/p95: 0/ 1/ 0.05/ 0.00 1112ms total, 17985.61 ops/sec
INCR, 50/5 min/max/avg/p95: 0/ 3/ 0.84/ 2.00 339ms total, 58997.05 ops/sec
LPUSH, 1/5 min/max/avg/p95: 0/ 1/ 0.05/ 1.00 1131ms total, 17683.47 ops/sec
LPUSH, 50/5 min/max/avg/p95: 0/ 3/ 0.86/ 2.00 345ms total, 57971.01 ops/sec
LRANGE 10, 1/5 min/max/avg/p95: 0/ 1/ 0.06/ 1.00 1228ms total, 16286.64 ops/sec
LRANGE 10, 50/5 min/max/avg/p95: 0/ 3/ 0.95/ 2.00 382ms total, 52356.02 ops/sec
LRANGE 100, 1/5 min/max/avg/p95: 0/ 1/ 0.08/ 1.00 1567ms total, 12763.24 ops/sec
LRANGE 100, 50/5 min/max/avg/p95: 0/ 6/ 1.68/ 3.00 675ms total, 29629.63 ops/sec
SET 4MiB buf, 1/5 min/max/avg/p95: 1/ 4/ 1.37/ 2.00 692ms total, 722.54 ops/sec
SET 4MiB buf, 50/5 min/max/avg/p95: 3/ 183/ 57.79/ 125.00 605ms total, 826.45 ops/sec
GET 4MiB str, 1/5 min/max/avg/p95: 5/ 16/ 8.14/ 12.95 816ms total, 122.55 ops/sec
GET 4MiB str, 50/5 min/max/avg/p95: 24/ 323/ 202.98/ 309.00 519ms total, 192.68 ops/sec
GET 4MiB buf, 1/5 min/max/avg/p95: 6/ 13/ 8.01/ 11.95 802ms total, 124.69 ops/sec
GET 4MiB buf, 50/5 min/max/avg/p95: 16/ 480/ 203.85/ 435.70 531ms total, 188.32 ops/sec

## TODO

Better tests for auth, disconnect/reconnect, and all combinations thereof.
Expand All @@ -674,7 +738,7 @@ I think there are more performance improvements left in there for smaller values
comment again with indignation!)

## Contributors
Some people have have added features and fixed bugs in `node_redis` other than me.
Many people have have added features and fixed bugs in `node_redis`.

Ordered by date of first contribution.
[Auto-generated](http://github.com/dtrejo/node-authors) on Wed Jul 25 2012 19:14:59 GMT-0700 (PDT).
Expand Down Expand Up @@ -740,5 +804,3 @@ HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
OTHER DEALINGS IN THE SOFTWARE.

![spacer](http://ranney.com/1px.gif)

0 comments on commit eb03bb9

Please sign in to comment.