How to choose an in-memory NoSQL solution: Performance measuring

The main purpose of this work is to show results of benchmarking some  of the leading in-memory NoSQL databases with a tool named YCSB.

We selected three popular in-memory database management systems: Redis (standalone and in-cloud named Azure Redis Cache), Tarantool and CouchBase and one cache system Memcached.  Memcached is not a database management system and does not have  persistence. But we decided to take it, because it is also widely used  as a fast storage system. Our “firing field” was a group of four virtual  machines in Microsoft Azure Cloud. Virtual machines are located close  to each other, meaning they are in one datacenter. This is necessary to  reduce the impact of network overhead in latency measurements. Images of these VMs can be downloaded by links: one, two, three and four (login: nosql, password: qwerty). A pair of VMs named nosql-1 and nosql-2 is useful for benchmarking Tarantool and CouchBase and another pair of VMs named nosql-3 and nosql-4 is good for Redis, Azure Redis Cache and Memcached. Databases and tests are installed and configured on these images.

Our virtual machines were the basic A3 instances with 4 cores, 7 GB RAM and 120 GB disk size.

Databases and their configurations

In-memory NoSQL database management system is a database management  system that stores all the data in the main memory and persists each  update on disk. Persistency is provided by saving each data modification request in a  binary log. Since the log is written in append-only mode, it is rarely a  bottleneck. Both read and write workloads are processed without significant disk  head movement.


Redis has been around since 2009 and created and supported by Salvatore Sanfilippo, the newest version of is 3.0.5.
We provided two configuration files for Redis 3.0.4 – with an append-only file for data persistence and without it like a cache server. Redis was built from source.

Azure Redis Cache

We also used Microsoft Azure Redis Cache. It is based on Redis cloud service managed by Microsoft.


Tarantool is an open-source NoSQL database management system and Lua application server developed in The first version of Tarantool was released in 2008 and the newest version is 1.6.7.
We provided four configuration files for Tarantool 1.6.7-126-gb35aff9 – both with write-ahead log and without it and both with tree (ordered)  and hash (unordered) indices. No tuning options exist in Tarantool and  no tuning was done.


Memcached is a general-purpose distributed memory caching system which has been developed in 2003.
Memcached does not have any mode with data persistence so it was tested  only with other data-bases which are not configured with any append-only  binary log. We used Memcached 1.4.14-0ubuntu9 from Ubuntu repository.


Couchbase Server, originally known as Membase, is an open-source,  distributed NoSQL document-oriented database. The newest stable release  is 4.0.
We used Couchbase 4.0.0-4047-1 package from official site without any extra configuration.

Append-only files in Redis and write-ahead logs in Tarantool options  enable data persistence for current databases. Comparisons only for  similar configurations of different databases are described in this  paper. It means we don’t compare, for example, Redis with enabled  append-only files and Tarantool with disabled write-ahead logs.

Yahoo! Cloud Serving Benchmark

Yahoo! Cloud Serving Benchmark,  or YCSB is a powerful utility for performance measuring of a wide range  of NoSQL databases including in-memory and on-disk solutions. YCSB is a  branch standard for performance measuring of NoSQL solutions, which is  why we are using it. We are interested in Redis and Tarantool drivers  which are included in YCSB and the Memcached driver which is created by  us based on the spymemcached library. The source of this YCSB branch can be seen here.

YCSB provides few core workload types that are presented in its own  directory as configuration files. There are six major workload types  named by letters from A to F.

Workload A is an update heavy workload. It has a mix of 50/50  reads and writes. An application example is a session store recording  recent actions. Workload B is primarily a read workload. It has a  95/5 read/write mix. Application example: photo tagging; add a tag is  an update, but most operations are to read tags. Workload C is 100% read. Application example: user profile cache, where profiles are constructed elsewhere (e.g., Hadoop).  In Workload D,  new records are inserted and the most recently inserted records are the  most popular. Application example: user status updates; people want to  read the latest. In Workload E, short ranges of records are  queried instead of individual records. Application example: threaded  conversations, where each scan is for the posts in a given thread  (assumed to be clustered by thread id). In Workload F, the client  will read a record, modify it, and write back the changes. Application  example: user database, where user records are read and modified by the  user or to record user activity.

We have changed two parameters in each of these configuration files: recordcount to 2000000 and operationcount to 5000000. YCSB is a multithreaded tester and we start it with 8, 16, 32, 64, 128 and 256 threads.

Now we will show and describe some packs of plots drawn by us in R. Sources of plot scrips can be downloaded here.


Tarantool (HASH)
Tarantool (TREE)
Azure Redis Cache

Tarantool with both hash and tree indices is the best for all  investigated workloads. It creates a lock-free in-memory engine, which  does not consist of any mutexes or other concurrency primitives and uses  cooperative multitasking. After considering these graphs, we can  conclude that high throughput is one of the strengths of the Tarantool  database.

The design of Tarantool shows the minimal average latency for read  requests too. As we can see on these plots, this is true for any  workload. On the 95% requests Tarantool reaches the lowest latency too.  (This measure is related to average latency but they are not the same).

However, on the 99% fastest requests, Tarantool does not reaches  lowest latency for any workload. By this measure, Tarantool is really  close to Redis in all cases and beaten by it in some of them. This  situation can be described as follows: Tarantool executes part of the  queries with a small latency and another part with a large latency,  while Redis executes all requests with a middle latency.

For cases without write-ahead logs, Memcached and Couchbase exhibit better latency.

In any case Tarantool is better than Redis by average latency and  95th percentile, but not by 99th percentile. This situation is similar  with read latency and can be described in a similar way.

On insert requests and on cases without write-ahead logs, Memcached  and Couchbase exhibit better results than others do in average latency  and 95th percentile, but in 99th percentile we can see a completely  opposite picture – these databases with Azure Redis Cache provide worse  results than all others, another fact we discovered that shows Tarantool  leading in comparison with Redis.

In this case (Read-Modify-Write), Tarantool again had the best result in average and 95th percentile and was very close to Redis in 99th percentile.


We described YCSB and have provided the results of comparing four  popular databases, but the most significant idea considered in this  paper is the way of choosing the right solution for the current  workload. By looking at the plots placed within this article, it is  simple to find the most suitable solution with respect to your workload  type, database clients count and your expectations.

The links on our VMs images, YCSB with Memcached module and R scripts  are specially published so that you can conduct your own tests and  verify our results or get results for instances of different  configurations (both hardware and software).

Through all tests we executed, Tarantool showed the best result for  the count requests per second and for many of tests latency values on  any type of examined workloads. Therefore, we can decide that for most  of typical projects Tarantool suits them more that popular solutions  such as Redis, CouchBase or Memcached. This is the basis of our decision  to use Tarantool for our projects here at

On HackerNews