We’ve been hard at work on TokuMX since it’s initial release just over 2 months ago. Today we released TokuMX v1.2 which includes Hot Backup in the Enterprise Edition.

Hot Backup allows users to create a backup of a running TokuMX primary or secondary server in a replica set, with no blocking of writes for clients. We will be blogging more about the Hot Backup technology in the coming weeks. This same technology is used for Hot Backup in TokuDB.

Also worth noting are the features we’ve added since the initial TokuMX release:

• Migration Tools. Migrate to TokuMX from MongoDB using our tool that replays MongoDB repication. This allows a TokuMX server to stay in sync with a MongoDB replica set, reducing downtime for production go-live.
• Bulk Loading. Significantly improves data load speeds …

Recently, we’ve seen a few people ask us about building TokuMX from scratch. While it’s best if you just use the binaries you can get from us (they have all the right optimizations, we’ve tested them, and we can interpret coredumps they generate), we recognize there are other reasons you might need to do a custom build.

Since we actually build six distinct products all using the Fractal Tree indexing® library (community and enterprise versions of TokuDB for MySQL, TokuDB for MariaDB, and TokuMX), our build process is pretty complicated, compared to software packages that might, for example, just involve one source repository and link against a few standard libraries. Our TokuMX builds involve four git repositories, three separate build stages, two different build tools, and …

On Wednesday night, the Boston MongoDB User group was kind enough to have me speak about TokuMX Internals. I spoke about Fractal Tree® indexes and the technical reasons behind the benefits they provide to MongoDB applications. Although the talk mostly references TokuMX and MongoDB, all the theory applies to TokuDB and MySQL as well.

My slides are on our technology overview page, along with other great content.

Opportunities to present technical material to an engaged audience asking tough questions is rare, and much appreciated. So thank you to the Boston MongoDB User group for having …

A lot is said about the differences in the data between MySQL and MongoDB. Things such as “MongoDB is document based”, “MySQL is relational”, “InnoDB has a clustering key”, etc.. Some may wonder how TokuDB, our MySQL storage engine, and TokuMX, our MongoDB product, fit in with these data layouts. I could not find anything describing the differences with a simple google search, so I figured I’d write a post explaining how things compare.

So who are the players here? With MySQL, users are likely familiar with two storage engines: MyISAM, the original default up until …

Before creating a unique index in TokuMX or TokuDB, ask yourself, “does my application really depend on the database enforcing uniqueness of this key?” If the answer is ANYTHING other than yes, do not declare the index to be unique. Why? Because unique indexes may kill your write performance. In this post, I’ll explain why.

Unique indexes are a strange beast: they have no impact on standard databases that use B-Trees, such as MongoDB and MySQL, but may be horribly painful for databases that use write optimized data structures, like TokuMX’s Fractal Tree(R) indexes. How? …

In my last post, I showed what a Fractal Tree® index is at a high level. Once again, the Fractal Tree index is the data structure inside TokuMX and TokuDB, our MongoDB and MySQL products. One of its strengths is the ability to get high levels of compression on the stored data. In this post, I’ll explain why that is.

At a high level, one can argue that there isn’t anything special about our compression algorithms. We basically do this: we take large chunks of data, use known compression methods (e.g. zlib, lzma, …

With our recent release of TokuMX 1.0, we’ve made some bold claims about how fast TokuMX can run MongoDB workloads. In this post, I want to dig into one of the big areas of improvement, write performance and reduced I/O.

One of the innovations of TokuMX is that it eliminates a long-held rule of databases: to get good write performance, the working set of your indexes should fit in memory. The standard reasoning goes along the lines of: if your indexes’ working set does not fit in memory, then your writes will induce I/O, you will become I/O bound, and performance will suffer. So, either make sure your indexes fit in …

I am actually quite excited about Tokutek’s release of TokuMX. I think it is going to change the landscape of database systems and it is finally something that made me looking into NoSQL.

Why is TokuMX interesting? A few reasons:

• It comes with transactions, and all that good stuff that transactions provide: a concurrent access to documents (no more global write-lock in MongoDB); crash recovery; atomicity
• Performance in IO-bound operations
• A good compression rate, which is a money-saver if you use SSD/Flash
• But it is also SSD/Flash life-time friendly, which is double money-saver

So having all these factors it is just a no-brainer if you have a …

Tokutek is known for its full-featured fast-indexing technology. MongoDB is known for its great document-based data model and ease of use. TokuMX, version 1.0, combines the best of both worlds.

• So what, exactly, is TokuMX? The simplest (but incomplete) answer is that TokuMX is MongoDB with all its storage code replaced by Tokutek’s Fractal Tree indexes.
• How do Fractal Tree indexes improve MongoDB? The direct benefits include high-performance indexing, strong compression, and performance stability – in other words, the performance stays high, even when data is larger than RAM.
• Are there any features in TokuMX that MongoDB doesn’t have? Yes. We have added support for transactions to TokuMX, so that TokuMX is ACID compliant and has MVCC. We have also added support for clustering indexes, which dramatically accelerate many types of queries.
• …

Tokutek created the iiBench benchmark back in 2008. The point of the benchmark is to measure the performance of indexed insertions over time. It uses an extremely simple schema, one table with a sequential insertion pattern for the primary key along with three integer fields storing random values. The table maintains 3 secondary indexes, each including several of the random integer fields. The iiBench application itself is currently maintained on Launchpad.

B-tree implementations generally require maintenance operations to update leaf nodes (an insertion is one such operation). When the entire B-tree index does not fit in RAM an IO is required, and performance drops dramatically. Fractal Tree Indexes do not exhibit this performance drop as leaf node …