6 ms·

MariaDB recently released the version 10.1 as GA. https://blog.mariadb.org/mariadb-10-1-is-stable-ga/ https://blog.mariadb.org/mariadb-10-1-is-stable-ga/ We b

by bratao 11y ago

MariaDB recently released the version 10.1 as GA. https://blog.mariadb.org/mariadb-10-1-is-stable-ga/ https://blog.mariadb.org/mariadb-10-1-is-stable-ga/ We been using it for a large dataset, and has been fantastic. Compared to MySQL 5.7 and PostgreSQL, it have the advantage of supporting the TokuDB engine out of box. My data uncompressed is 3TB, with it , we can fit in 300GB with all indexes. Read Free Replication with TokuDB (https://github.com/percona/tokudb-engine/wiki/Read-Free-Replication-with-TokuDB https://github.com/percona/tokudb-engine/wiki/Read-Free-Repl...) also enable us to have a very cheap VPS as slave.

swasheck 11y ago
could you explain why the tokudb engine is an advantage?
- cies 11y ago
  I found this: https://en.wikipedia.org/wiki/TokuDB https://en.wikipedia.org/wiki/TokuDB Fractal tree indexes[1]... Dude! 1: https://en.wikipedia.org/wiki/Fractal_tree_index https://en.wikipedia.org/wiki/Fractal_tree_index
- bratao 11y ago
  TokuDB is a very modern engine that uses a data-structure called "Fractal Trees". Imagine something like a B-Tree with buffers in the non-leaf nodes. It also heavily compress the row data. The main point of TokuDB is not writing everything to the disk thanks to the buffers, and with the very efficient compression, more data fits in the memory and saves the space for big-data applications. ( please someone correct me if I´m wrong) In our tests TokuDB was vital for our startup. With it, we can use cheap dedicated servers and our performance is amazing. We tried PostgreSQL, MongoDB 3.0 and MySQL 5.7 and they can´t fit our data in a 2TB disk or were slow in our tests.
  - tomcam 11y ago
    Not writing everything to disk ASAP makes me... queasy...
    - ddorian43 11y ago
      The log is always written, what's not written asap is changes on the tree (changes go into the buffers first and then trickle down when buffers are full).
      - leif 11y ago
        This is the critical insight. TokuDB uses a write-ahead log which is synced according to the configuration, and can be made as immediate as full fsync on commit. This provides the strongest durability available on a single machine. Where TokuDB gets its speed boost is by delaying the random reads associated with updating the indexing structure (the Fractal Tree). The buffers are written to disk on checkpoint, but because they're buffers, the potentially random writes are localized to a smaller number of nodes high in the tree, which minimizes the number of disk seeks required. Since sequential I/O is cheaper than random, the sequential writes to the write-ahead log are very fast, so even in very strict durability configurations, TokuDB can easily outperform databases which use random writes to update the indexing structures, such as the B-trees used by InnoDB and most other RDBMSes. More details here: https://www.percona.com/blog/2011/09/22/write-optimization-myths-comparison-clarifications/ https://www.percona.com/blog/2011/09/22/write-optimization-m...
    - rch 11y ago
      It's fine for people who use RDBMSs for data processing, like loading daily log data into tables and running one or more massive queries to yield averages, etc. etc. Not ideal, but simple and valid enough.
      - nickpsecurity 11y ago
        It's fine for regular loads, too, so long as they do regular saves to disk and have a high availability setup. Plenty of setups out there using in-memory, checkpointed DB's and RAM drives on reliable hardware without serious problems.
- acomjean 11y ago
  Tokutek and some staff where common vititors to the boston mysql meetup. If I remember they were using "fractal trees" instead of b-trees. Trading cpu cycles for disk io. It was interesting tech. http://highscalability.com/blog/2014/8/6/tokutek-white-paper-a-comparison-of-log-structured-merge-lsm.html http://highscalability.com/blog/2014/8/6/tokutek-white-paper... Quote:"The idea behind FT(fractal Trees) indexes is to maintain a B tree in which each internal node of the tree contains a buffer. When a data record is inserted into the tree, instead of traversing the entire tree the way a B tree would, we simply insert the Eventually the root buffer will fill up with new data records. At that point the FT index copies the inserted records down a level of the tree. Eventually the newly inserted records will reach the leaves, at which point they are simply stored in a leaf node as a B tree would store them. The data records descending through the buffers of the tree can be thought of as messages that say “insert this record”. FT indexes can use other kinds of messages, such as messages that delete a record, or messages that update a record." some other links: http://web.archive.org/web/20150414215556/http://www.tokutek.com/2010/06/making-replace-into-fast-by-avoiding-disk-seeks/ http://web.archive.org/web/20150414215556/http://www.tokutek... http://assets.en.oreilly.com/1/event/36/How%20TokuDB%20Fractal%20Tree%20Databases%20Work%20Presentation.pdf http://assets.en.oreilly.com/1/event/36/How%20TokuDB%20Fract...
  - bcoates 11y ago
    Are there any comparisons between fractal trees and the database cracking/adaptive indices advocated by [1]? It seems like they're going after the same use pattern but cracking is relatively simpler and even faster for inserts. [1] http://db.csail.mit.edu/pubs/abadi-column-stores.pdf http://db.csail.mit.edu/pubs/abadi-column-stores.pdf
    - acomjean 11y ago
      Not that I know of. Interesting though. I have to look into column stores more though.
    - leif 11y ago
      Cracking is kind of similar in that it delays the "sorting work" done in the indexing structure. However, cracking is fairly heuristic and therefore hard to analyze without an intimate understanding of the workload. Fractal Trees do pretty much the same thing under all workloads (modulo optimizations for things like sequential and Zipfian workloads), so they're easier to analyze and prove bounds for. An interesting new development is http://getkudu.io/ http://getkudu.io/ which applies some ideas common with Fractal Trees and other write-optimized data structures (like LSM-trees) to column stores. At Tokutek, we had some designs for how to implement column store-like data models on Fractal Trees (we called them "matrix stores" because they had some of the benefits of both row stores and column stores), but we didn't get around to implementing them.
- nekitamo 11y ago
  It has very high insert speeds, and amazing compression (up to 500% in my tests). It's very powerful if you're inserting terabytes of data every day on commodity hardware.
- sjwright 11y ago
  For our workload (discussion forum, custom engine) the move to TokuDB was so effective that we ended up completely eliminating a number of front-end caching stages, because rendering the page from scratch every time ended up being very nearly as fast as pulling from the cache. (The hot path was a single well-optimised query that retrieved all content for the page. The query didn't need to change.)
  - mooreds 11y ago
    Wow, that's a great datapoint.
    - sjwright 11y ago
      To be more precise, the cache was still faster than building on the fly, but given the quantity of content we were dealing with, the cache was a resource hog. Were it not for TokuDB our next move probably would have been to deploy a dedicated cache server with huge amounts of RAM. As it stands now, this content is now being handled by a single (and quite unremarkable) MariaDB slave and assembled on-the-fly like any other web page. For anyone dealing with gigabytes of text content, I highly recommend giving it a go. You don't even need to screw with your master, just change the table type on the slave.
      - Jgrubb 11y ago
        Hey if you're still paying attention to this thread I'd love to pick your brain about this topic. contact in my profile if you don't mind. Thanks.
- mathnode 11y ago
  My favourite, TokuDB has online DDL. Which is a welcome change to the MySQL ecosystem. I am pretty sure the only other engines that support online DDL is Sybase and DB2. Online DDL generally means lock free DDL operations. Not even Oracle, MSSQL, or Postgresql do online DDL. But at least they have transactional DDL, even if it isn't entirely lock-free all the time. Compared to InnoDB/XtraDB; TokuDB lacks foreign key support, because this was worked into InnoDB years ago because the MySQL server layer doesn't support any kind of constraints. I am hoping for MariaDB to start implementing more modern SQL features which would this be available in supported storage engines.
  - coolgeek 11y ago
    https://dev.mysql.com/doc/refman/5.6/en/innodb-online-ddl.html https://dev.mysql.com/doc/refman/5.6/en/innodb-online-ddl.ht...
    - leif 11y ago
      That's not really online, it still rebuilds the entire table, it just does it in the background and shows the results once it's done. This is pretty similar to how TokuDB hot indexing works, but TokuDB's hot column add/rename/delete is truly online, the results are visible immediately and the actual disk work is done in a lazy fashion.
imakesnowflakes 11y ago
Can I now refer to the same temporary table multiple times in a query?
allan_s 11y ago
About TokuDB and PostgreSQL How does one ,by tunning a vanilla PostgreSQL (on a per instance or per database or per table basis) can get the same kind of advantages (and tradeoff) than in MySQL/MariaDB by switching from InnoDB to TokuDB ?
- mathnode 11y ago
  Which aspect in particular do you wish to emulate from TokuDB in PostgreSQL?
  - ak4g 11y ago
    That compression ratio is one I'd love to have. PG's built-in data compression is limited to large values in a single row - if you have a hundred billion 100-byte tuples, TOAST isn't going to help. The closest thing I know of to achieve something comparable with PG would be to use a data directory mounted on something with transparent filesystem-level compression (meaning, practically speaking, ZFS). This gives you the less-than-ideal choice of a not-mainline-linux filesystem (for your database's data directory, which is worth being nervous about) or running an OpenSolaris descendant, which is a big departure for plenty of people who have only ever run production dbs on Linux servers.
    - brobinson 11y ago
      I've used PostgreSQL with BTRFS + LZO compression on Linux. After I set it up, our large queries (2 minutes or > runtime) saw a speedup of 5-10x and our TPS went up by around 25%. BTRFS has been in the stable kernel for over two years, so it might be worth checking out!
      - agopaul 11y ago
        Mechanical or flash storage? Compression could give advantage on mechanical disks due to the reduced time spent on seeking, but I'm curious to know how they perform on SSD for which seek time is reduced to a minimum if not 0.
sown 11y ago
Do you use a lot of joins, writes along with the queries? I'm curious as to what might happen with other profiles.
_Codemonkeyism 11y ago
On the Postgres list "Sadly, fractal trees are an invention of Tokutek and are heavily and publically patented."

by bratao 11y ago

swasheck 11y ago

could you explain why the tokudb engine is an advantage?

cies 11y ago
I found this: https://en.wikipedia.org/wiki/TokuDB https://en.wikipedia.org/wiki/TokuDB Fractal tree indexes[1]... Dude! 1: https://en.wikipedia.org/wiki/Fractal_tree_index https://en.wikipedia.org/wiki/Fractal_tree_index
bratao 11y ago
TokuDB is a very modern engine that uses a data-structure called "Fractal Trees". Imagine something like a B-Tree with buffers in the non-leaf nodes. It also heavily compress the row data. The main point of TokuDB is not writing everything to the disk thanks to the buffers, and with the very efficient compression, more data fits in the memory and saves the space for big-data applications. ( please someone correct me if I´m wrong) In our tests TokuDB was vital for our startup. With it, we can use cheap dedicated servers and our performance is amazing. We tried PostgreSQL, MongoDB 3.0 and MySQL 5.7 and they can´t fit our data in a 2TB disk or were slow in our tests.
- tomcam 11y ago
  Not writing everything to disk ASAP makes me... queasy...
  - ddorian43 11y ago
    The log is always written, what's not written asap is changes on the tree (changes go into the buffers first and then trickle down when buffers are full).
    - leif 11y ago
      This is the critical insight. TokuDB uses a write-ahead log which is synced according to the configuration, and can be made as immediate as full fsync on commit. This provides the strongest durability available on a single machine. Where TokuDB gets its speed boost is by delaying the random reads associated with updating the indexing structure (the Fractal Tree). The buffers are written to disk on checkpoint, but because they're buffers, the potentially random writes are localized to a smaller number of nodes high in the tree, which minimizes the number of disk seeks required. Since sequential I/O is cheaper than random, the sequential writes to the write-ahead log are very fast, so even in very strict durability configurations, TokuDB can easily outperform databases which use random writes to update the indexing structures, such as the B-trees used by InnoDB and most other RDBMSes. More details here: https://www.percona.com/blog/2011/09/22/write-optimization-myths-comparison-clarifications/ https://www.percona.com/blog/2011/09/22/write-optimization-m...
  - rch 11y ago
    It's fine for people who use RDBMSs for data processing, like loading daily log data into tables and running one or more massive queries to yield averages, etc. etc. Not ideal, but simple and valid enough.
    - nickpsecurity 11y ago
      It's fine for regular loads, too, so long as they do regular saves to disk and have a high availability setup. Plenty of setups out there using in-memory, checkpointed DB's and RAM drives on reliable hardware without serious problems.
acomjean 11y ago
Tokutek and some staff where common vititors to the boston mysql meetup. If I remember they were using "fractal trees" instead of b-trees. Trading cpu cycles for disk io. It was interesting tech. http://highscalability.com/blog/2014/8/6/tokutek-white-paper-a-comparison-of-log-structured-merge-lsm.html http://highscalability.com/blog/2014/8/6/tokutek-white-paper... Quote:"The idea behind FT(fractal Trees) indexes is to maintain a B tree in which each internal node of the tree contains a buffer. When a data record is inserted into the tree, instead of traversing the entire tree the way a B tree would, we simply insert the Eventually the root buffer will fill up with new data records. At that point the FT index copies the inserted records down a level of the tree. Eventually the newly inserted records will reach the leaves, at which point they are simply stored in a leaf node as a B tree would store them. The data records descending through the buffers of the tree can be thought of as messages that say “insert this record”. FT indexes can use other kinds of messages, such as messages that delete a record, or messages that update a record." some other links: http://web.archive.org/web/20150414215556/http://www.tokutek.com/2010/06/making-replace-into-fast-by-avoiding-disk-seeks/ http://web.archive.org/web/20150414215556/http://www.tokutek... http://assets.en.oreilly.com/1/event/36/How%20TokuDB%20Fractal%20Tree%20Databases%20Work%20Presentation.pdf http://assets.en.oreilly.com/1/event/36/How%20TokuDB%20Fract...
- bcoates 11y ago
  Are there any comparisons between fractal trees and the database cracking/adaptive indices advocated by [1]? It seems like they're going after the same use pattern but cracking is relatively simpler and even faster for inserts. [1] http://db.csail.mit.edu/pubs/abadi-column-stores.pdf http://db.csail.mit.edu/pubs/abadi-column-stores.pdf
  - acomjean 11y ago
    Not that I know of. Interesting though. I have to look into column stores more though.
  - leif 11y ago
    Cracking is kind of similar in that it delays the "sorting work" done in the indexing structure. However, cracking is fairly heuristic and therefore hard to analyze without an intimate understanding of the workload. Fractal Trees do pretty much the same thing under all workloads (modulo optimizations for things like sequential and Zipfian workloads), so they're easier to analyze and prove bounds for. An interesting new development is http://getkudu.io/ http://getkudu.io/ which applies some ideas common with Fractal Trees and other write-optimized data structures (like LSM-trees) to column stores. At Tokutek, we had some designs for how to implement column store-like data models on Fractal Trees (we called them "matrix stores" because they had some of the benefits of both row stores and column stores), but we didn't get around to implementing them.
nekitamo 11y ago
It has very high insert speeds, and amazing compression (up to 500% in my tests). It's very powerful if you're inserting terabytes of data every day on commodity hardware.
sjwright 11y ago
For our workload (discussion forum, custom engine) the move to TokuDB was so effective that we ended up completely eliminating a number of front-end caching stages, because rendering the page from scratch every time ended up being very nearly as fast as pulling from the cache. (The hot path was a single well-optimised query that retrieved all content for the page. The query didn't need to change.)
- mooreds 11y ago
  Wow, that's a great datapoint.
  - sjwright 11y ago
    To be more precise, the cache was still faster than building on the fly, but given the quantity of content we were dealing with, the cache was a resource hog. Were it not for TokuDB our next move probably would have been to deploy a dedicated cache server with huge amounts of RAM. As it stands now, this content is now being handled by a single (and quite unremarkable) MariaDB slave and assembled on-the-fly like any other web page. For anyone dealing with gigabytes of text content, I highly recommend giving it a go. You don't even need to screw with your master, just change the table type on the slave.
    - Jgrubb 11y ago
      Hey if you're still paying attention to this thread I'd love to pick your brain about this topic. contact in my profile if you don't mind. Thanks.
mathnode 11y ago
My favourite, TokuDB has online DDL. Which is a welcome change to the MySQL ecosystem. I am pretty sure the only other engines that support online DDL is Sybase and DB2. Online DDL generally means lock free DDL operations. Not even Oracle, MSSQL, or Postgresql do online DDL. But at least they have transactional DDL, even if it isn't entirely lock-free all the time. Compared to InnoDB/XtraDB; TokuDB lacks foreign key support, because this was worked into InnoDB years ago because the MySQL server layer doesn't support any kind of constraints. I am hoping for MariaDB to start implementing more modern SQL features which would this be available in supported storage engines.
- coolgeek 11y ago
  https://dev.mysql.com/doc/refman/5.6/en/innodb-online-ddl.html https://dev.mysql.com/doc/refman/5.6/en/innodb-online-ddl.ht...
  - leif 11y ago
    That's not really online, it still rebuilds the entire table, it just does it in the background and shows the results once it's done. This is pretty similar to how TokuDB hot indexing works, but TokuDB's hot column add/rename/delete is truly online, the results are visible immediately and the actual disk work is done in a lazy fashion.

imakesnowflakes 11y ago

Can I now refer to the same temporary table multiple times in a query?

allan_s 11y ago

About TokuDB and PostgreSQL How does one ,by tunning a vanilla PostgreSQL (on a per instance or per database or per table basis) can get the same kind of advantages (and tradeoff) than in MySQL/MariaDB by switching from InnoDB to TokuDB ?

mathnode 11y ago
Which aspect in particular do you wish to emulate from TokuDB in PostgreSQL?
- ak4g 11y ago
  That compression ratio is one I'd love to have. PG's built-in data compression is limited to large values in a single row - if you have a hundred billion 100-byte tuples, TOAST isn't going to help. The closest thing I know of to achieve something comparable with PG would be to use a data directory mounted on something with transparent filesystem-level compression (meaning, practically speaking, ZFS). This gives you the less-than-ideal choice of a not-mainline-linux filesystem (for your database's data directory, which is worth being nervous about) or running an OpenSolaris descendant, which is a big departure for plenty of people who have only ever run production dbs on Linux servers.
  - brobinson 11y ago
    I've used PostgreSQL with BTRFS + LZO compression on Linux. After I set it up, our large queries (2 minutes or > runtime) saw a speedup of 5-10x and our TPS went up by around 25%. BTRFS has been in the stable kernel for over two years, so it might be worth checking out!
    - agopaul 11y ago
      Mechanical or flash storage? Compression could give advantage on mechanical disks due to the reduced time spent on seeking, but I'm curious to know how they perform on SSD for which seek time is reduced to a minimum if not 0.

sown 11y ago

Do you use a lot of joins, writes along with the queries? I'm curious as to what might happen with other profiles.

_Codemonkeyism 11y ago

On the Postgres list "Sadly, fractal trees are an invention of Tokutek and are heavily and publically patented."