Indexing speed: web v.s. solrj app

[Lance Norskog <go...@gmail.com>]

Is indexing via solrj faster than going through the web service? There are
three cases:
    Read a file from a local file system and indexing it directly,
    Read a file on one machine and indexing it on another, and
    Run solrj and read a file, then directly update the index.

I'm talking about the last case v.s. the middle case: we are already feeding
solr from remote machines.

Thanks,

Lance

Re: Indexing speed: web v.s. solrj app

[Yonik Seeley <yo...@apache.org>]

On 8/15/07, Lance Norskog <go...@gmail.com> wrote:
> Is indexing via solrj faster than going through the web service? There are
> three cases:
>     Read a file from a local file system and indexing it directly,
>     Read a file on one machine and indexing it on another, and
>     Run solrj and read a file, then directly update the index.
>
> I'm talking about the last case v.s. the middle case: we are already feeding
> solr from remote machines.

solrj is (in my opinion) primarily as a convenience client for going
through the web service.
If you mean using solrj locally, I would guess that giving the CSV
handler a local file (or perhaps even streaming remotely) might be a
tad faster.

If done correctly, the differences can be small enough that
architectural decisions shouldn't be based on those differences.

-Yonik