Dragonfly is a modern in-memory datastore, fully compatible with Redis and Memcached APIs. Dragonfly implements novel algorithms and data structures on top of a multi-threaded, shared-nothing architecture. As a result, Dragonfly reaches x25 performance compared to Redis and supports millions of QPS on a single instance.
Dragonfly's core properties make it a cost-effective, high-performing, and easy-to-use Redis replacement.
Dragonfly is crossing 3.8M QPS on c6gn.16xlarge reaching x25 increase in throughput compared to Redis.
99th latency percentile of Dragonfly at its peak throughput:
op | r6g | c6gn | c7g |
---|---|---|---|
set | 0.8ms | 1ms | 1ms |
get | 0.9ms | 0.9ms | 0.8ms |
setex | 0.9ms | 1.1ms | 1.3ms |
All benchmarks were performed using memtier_benchmark
(see below) with number of threads tuned per server type and the instance type. memtier
was running on a separate c6gn.16xlarge machine. For setex benchmark we used expiry-range of 500, so it would survive the end of the test.
memtier_benchmark --ratio ... -t <threads> -c 30 -n 200000 --distinct-client-seed -d 256 \
--expiry-range=...
When running in pipeline mode --pipeline=30
, Dragonfly reaches 10M qps for SET and 15M qps for GET operations.
In the following test, we filled Dragonfly and Redis with ~5GB of data
using debug populate 5000000 key 1024
command. Then we started sending the update traffic with memtier
and kicked off the snapshotting with the
"bgsave" command. The following figure demonstrates clearly how both servers behave in terms of memory efficiency.
Dragonfly was 30% more memory efficient than Redis at the idle state. It also did not show any visible memory increase during the snapshot phase. Meanwhile, Redis reached almost x3 memory increase at peak compared to Dragonfly. Dragonfly also finished the snapshot much faster, just a few seconds after it started. For more info about memory efficiency in Dragonfly see dashtable doc
Dragonfly runs on linux. It uses relatively new linux specific io-uring API for I/O, hence it requires Linux version 5.10 or later. Debian/Bullseye, Ubuntu 20.04.4 or later fit these requirements.
docker pull docker.dragonflydb.io/dragonflydb/dragonfly && \
docker tag docker.dragonflydb.io/dragonflydb/dragonfly dragonfly
docker run --network=host --ulimit memlock=-1 --rm dragonfly
redis-cli PING # redis-cli can be installed with "apt install -y redis-tools"
You need --ulimit memlock=-1
because some Linux distros configure the default memlock limit for containers as 64m and Dragonfly requires more.
We maintain binary releases for x86 and arm64 architectures. You will need to install libunwind8
lib to run the binaries.
You need to install dependencies in order to build on Ubuntu 20.04 or later:
git clone --recursive https://github.com/dragonflydb/dragonfly && cd dragonfly
# to install dependencies
sudo apt install ninja-build libunwind-dev libboost-fiber-dev libssl-dev \
autoconf-archive libtool
# Configure the build
./helio/blaze.sh -release
# Build
cd build-opt && ninja dragonfly
# Run
./dragonfly --alsologtostderr
Dragonfly supports common redis arguments where applicable.
For example, you can run: dragonfly --requirepass=foo --bind localhost
.
Dragonfly currently supports the following Redis-specific arguments:
port
bind
requirepass
maxmemory
dir
- by default, dragonfly docker uses/data
folder for snapshotting. You can use-v
docker option to map it to your host folder.dbfilename
In addition, it has Dragonfly specific arguments options:
memcache_port
- to enable memcached compatible API on this port. Disabled by default.keys_output_limit
- maximum number of returned keys inkeys
command. Default is 8192.keys
is a dangerous command. we truncate its result to avoid blowup in memory when fetching too many keys.dbnum
- maximum number of supported databases forselect
.cache_mode
- see Cache section below.
for more options like logs management or tls support, run dragonfly --help
.
Currently Dragonfly supports ~130 Redis commands and all memcache commands besides cas
.
We are almost on par with Redis 2.8 API. Our first milestone will be to stabilize basic
functionality and reach API parity with Redis 2.8 and Memcached APIs.
If you see that a command you need, is not implemented yet, please open an issue.
The next milestone will be implementing H/A with redis -> dragonfly
and
dragonfly<->dragonfly
replication.
For dragonfly-native replication we are planning to design a distributed log format that will support order of magnitude higher speeds when replicating.
After replication and failover feature we will continue with other Redis commands from APIs 3,4 and 5.
API 1.0
- String family
- SET
- SETNX
- GET
- DECR
- INCR
- DECRBY
- GETSET
- INCRBY
- MGET
- MSET
- MSETNX
- SUBSTR
- Generic family
- DEL
- ECHO
- EXISTS
- EXPIRE
- EXPIREAT
- KEYS
- PING
- RENAME
- RENAMENX
- SELECT
- TTL
- TYPE
- SORT
- Server Family
- AUTH
- QUIT
- DBSIZE
- BGSAVE
- SAVE
- DEBUG
- EXEC
- FLUSHALL
- FLUSHDB
- INFO
- MULTI
- SHUTDOWN
- LASTSAVE
- SLAVEOF/REPLICAOF
- SYNC
- Set Family
- SADD
- SCARD
- SDIFF
- SDIFFSTORE
- SINTER
- SINTERSTORE
- SISMEMBER
- SMOVE
- SPOP
- SRANDMEMBER
- SREM
- SMEMBERS
- SUNION
- SUNIONSTORE
- List Family
- LINDEX
- LLEN
- LPOP
- LPUSH
- LRANGE
- LREM
- LSET
- LTRIM
- RPOP
- RPOPLPUSH
- RPUSH
- SortedSet Family
- ZADD
- ZCARD
- ZINCRBY
- ZRANGE
- ZRANGEBYSCORE
- ZREM
- ZREMRANGEBYSCORE
- ZREVRANGE
- ZSCORE
- Other
- BGREWRITEAOF
- MONITOR
- RANDOMKEY
- MOVE
API 2.0
- List Family
- BLPOP
- BRPOP
- BRPOPLPUSH
- LINSERT
- LPUSHX
- RPUSHX
- String Family
- SETEX
- APPEND
- PREPEND (dragonfly specific)
- BITCOUNT
- BITFIELD
- BITOP
- BITPOS
- GETBIT
- GETRANGE
- INCRBYFLOAT
- PSETEX
- SETBIT
- SETRANGE
- STRLEN
- HashSet Family
- HSET
- HMSET
- HDEL
- HEXISTS
- HGET
- HMGET
- HLEN
- HINCRBY
- HINCRBYFLOAT
- HGETALL
- HKEYS
- HSETNX
- HVALS
- HSCAN
- PubSub family
- PUBLISH
- PUBSUB
- PUBSUB CHANNELS
- SUBSCRIBE
- UNSUBSCRIBE
- PSUBSCRIBE
- PUNSUBSCRIBE
- Server Family
- WATCH
- UNWATCH
- DISCARD
- CLIENT LIST/SETNAME
- CLIENT KILL/UNPAUSE/PAUSE/GETNAME/REPLY/TRACKINGINFO
- COMMAND
- COMMAND COUNT
- COMMAND GETKEYS/INFO
- CONFIG GET/REWRITE/SET/RESETSTAT
- MIGRATE
- ROLE
- SLOWLOG
- PSYNC
- TIME
- LATENCY...
- Generic Family
- SCAN
- PEXPIREAT
- PEXPIRE
- DUMP
- EVAL
- EVALSHA
- OBJECT
- PERSIST
- PTTL
- RESTORE
- SCRIPT LOAD/EXISTS
- SCRIPT DEBUG/KILL/FLUSH
- Set Family
- SSCAN
- Sorted Set Family
- ZCOUNT
- ZINTERSTORE
- ZLEXCOUNT
- ZRANGEBYLEX
- ZRANK
- ZREMRANGEBYLEX
- ZREMRANGEBYRANK
- ZREVRANGEBYSCORE
- ZREVRANK
- ZUNIONSTORE
- ZSCAN
- HYPERLOGLOG Family
- PFADD
- PFCOUNT
- PFMERGE
Memchache API
- set
- get
- replace
- add
- stats (partial)
- append
- prepend
- delete
- flush_all
- incr
- decr
- version
- quit
Some commands were implemented as decorators along the way:
- ROLE (2.8) decorator as master.
- UNLINK (4.0) decorator for DEL command
- BGSAVE (decorator for save)
- FUNCTION FLUSH (does nothing)
Implement leader/follower replication (PSYNC/REPLICAOF/...).
APIs 3,4,5 without cluster support, without modules and without memory introspection commands. Also without geo commands and without support for keyspace notifications, without streams. Probably design config support. Overall - few dozens commands... Probably implement cluster-API decorators to allow cluster-configured clients to connect to a single instance.
Dragonfly has a single unified adaptive caching algorithm that is very simple and memory efficient.
You can enable caching mode by passing --cache_mode=true
flag. Once this mode
is on, Dragonfly will evict items least likely to be stumbled upon in the future but only when
it is near maxmemory limit.
Expiration ranges are limited to ~4 years. Moreover, expiration deadlines with millisecond precision (PEXPIRE/PSETEX etc) will be rounded to closest second for deadlines greater than 134217727ms (approximately 37 hours). Such rounding has less than 0.001% error which I hope is acceptable for large ranges. If it breaks your use-cases - talk to me or open an issue and explain your case.
For more detailed differences between this and Redis implementations see here.
By default Dragonfly allows http access via its main TCP port (6379). That's right, you
can connect to Dragonfly via Redis protocol and via HTTP protocol - the server recognizes
the protocol automatically during the connection initiation. Go ahead and try it with your browser.
Right now it does not have much info but in the future we are planning to add there useful
debugging and management info. If you go to :6379/metrics
url you will see some prometheus
compatible metrics.
Important! Http console is meant to be accessed within a safe network.
If you expose Dragonfly's TCP port externally, it is advised to disable the console
with --http_admin_console=false
or --nohttp_admin_console
.
Dragonfly started as an experiment to see how an in-memory datastore could look like if it was designed in 2022. Based on lessons learned from our experience as users of memory stores and as engineers who worked for cloud companies, we knew that we need to preserve two key properties for Dragonfly: a) to provide atomicity guarantees for all its operations, and b) to guarantee low, sub-millisecond latency over very high throughput.
Our first challenge was how to fully utilize CPU, memory, and i/o resources using servers that are available today in public clouds. To solve this, we used shared-nothing architecture, which allows us to partition the keyspace of the memory store between threads, so that each thread would manage its own slice of dictionary data. We call these slices - shards. The library that powers thread and I/O management for shared-nothing architecture is open-sourced here.
To provide atomicity guarantees for multi-key operations, we used the advancements from recent academic research. We chose the paper "VLL: a lock manager redesign for main memory database systems” to develop the transactional framework for Dragonfly. The choice of shared-nothing architecture and VLL allowed us to compose atomic multi-key operations without using mutexes or spinlocks. This was a major milestone for our PoC and its performance stood out from other commercial and open-source solutions.
Our second challenge was to engineer more efficient data structures for the new store. To achieve this goal, we based our core hashtable structure on paper "Dash: Scalable Hashing on Persistent Memory". The paper itself is centered around persistent memory domain and is not directly related to main-memory stores. Nevertheless, its very much applicable for our problem. It suggested a hashtable design that allowed us to maintain two special properties that are present in the Redis dictionary: a) its incremental hashing ability during datastore growth b) its ability to traverse the dictionary under changes using a stateless scan operation. Besides these 2 properties, Dash is much more efficient in CPU and memory. By leveraging Dash's design, we were able to innovate further with the following features:
- Efficient record expiry for TTL records.
- A novel cache eviction algorithm that achieves higher hit rates than other caching strategies like LRU and LFU with zero memory overhead.
- A novel fork-less snapshotting algorithm.
After we built the foundation for Dragonfly and we were happy with its performance, we went on to implement the Redis and Memcached functionality. By now, we have implemented ~130 Redis commands (equivalent to v2.8) and 13 Memcached commands.
And finally,
Our mission is to build a well-designed, ultra-fast, cost-efficient in-memory datastore for cloud workloads that takes advantage of the latest hardware advancements. We intend to address the pain points of current solutions while preserving their product APIs and propositions.