Memory database memsql--based on memory, mvcc+ hash table, skip tables

The hottest topic in the database industry this week is memsql, whether "old bottles of new wine" has sparked a lot of debate, and has raised questions about whether product technology is important or DBA. There are some introductions to memsql on the web, mostly from official documents. In this article, Curt Monash, a well-known independent analyst in the database industry, has also published his views on Memsql.

What the hell is Memsql?
In-Memory relational database
QL-92 subset compatible with MySQL (except for SQL coverage issues)

Performance of Memsql
Reading performance than memcached difference of about 10%
Write performance is about 20% stronger than memcached.
A 64-core, TBS-TB machine can run up to 1.2 million inserts per second
Under the same conditions, 500 million records can be loaded within 20 minutes.

The discussion about Memsql focuses on performance, including: data is organized by a hash table and a jump table (skip lists) . Memsql think the jump table on multicore extensibility very good query mode can be compiled into C + + mvcc/non-read lock lightweight write lock can adjust the persistence, you can fully durable run memsql, you can also set a buffer Size to limit the amount of transaction data that can be sustained
In fact, pre-compiling queries is hardly an innovation, nor is it a unique part of the industry. Earlier, including Qliktech, Streambase and Paraccel have done similar attempts. And Memsql is characterized by: compiled as long as you want to read the C + + parameterization, if a query includes a series of parameters, then it will be stored for future run more parameter persistence, if the server is down, the stored compilation query will not lose only a few kilobytes per query space, Of the early Memsql customers, they stored no more than thousands of query modes. So Memsql is very optimistic about the impact of compiling these queries, and the way that the LRU algorithm is used to free up space may not be considered.

Memsql Persistence uses a pre-write log to a disk (traditional or SSD) while sending a snapshot to another disk. The persistence design is continuous, but it is unclear whether it is in the full-durability scene.

Some other notable technical details include: The Memsql can be run in a multi-temperature environment, with DDL manual control. That is, the newer data will be put into the memsql, the old data into MySQL one point I think is the best practice, the Memsql team also uphold this concept, is to be confirmed in two or more server RAM, the first time to commit to write parallel group by, This Memsql team prides itself on MEMSQL data compression, which they consider to require data compression memsql in the OLAP payload is very high, so their target audience is those who have frequent transactions in the system users

TechTarget China original content, original link: http://www.searchdatabase.com.cn/showcontent_62951.htm

Memory Database memsql--memory-based, mvcc+ Hashtable, skip table