Benchmark results of Couchbase Lite vs SQLite (1)

[Szymon C]

Hi Guys,

/* Im reposting it with correct link */

I have found this document with benchmark results of Couchbase Lite performance in comparsion to SQLite.

http://www.cs.washington.edu/education/courses/cse544/13sp/final-projects/p19-tperrier.pdf

The difference is huge.

So I am curious about the feature plans and your opinion about the benchmark.

[Jens Alfke]

On Aug 5, 2013, at 12:29 AM, Szymon C <szy...@gmail.com> wrote:

I have found this document with benchmark results of Couchbase Lite performance in comparsion to SQLite.

http://www.cs.washington.edu/education/courses/cse544/13sp/final-projects/p19-tperrier.pdf

The difference is huge.

So I am curious about the feature plans and your opinion about the benchmark.

They’re comparing apples and oranges.

First of all, Couchbase Lite is implemented on top of SQLite, so it’s not really any kind of competitor to it (and could hardly be expected to be faster!)

Second, the reasons you’d want to use Couchbase Lite are (a) replication and (b) unstructured/schemaless data, neither of which SQLite offers.

The result is sort of like benchmarking an XML parser against raw fread/fwrite, finding that OMG it’s slower, and recommending people use raw I/O for their data files. (Or for that matter, benchmarking CoreData against the raw sqlite API.)

(Which is not to say that we can’t optimize CBL further. Query speed has already been improved a lot over TouchDB. It’s been quite a while since I’ve gotten any reports of slow performance, and without concrete use cases to profile it’s hard to know what optimizations to focus on.)

—Jens