News for hackers. Stuff that matters.
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nzadrozny on Tuesday October 11, 2011 @05:50PM (#3100459)
Congrats to the LinkedIn and IndexTank folks for the successful exit/acquisition.

Apologies for tooting my own horn here, but for anyone wondering about alternatives for hosted search, there's always Apache Solr and fizx's and my hosted service at http://websolr.com/. We're bootstrapped, profitable, and going on two years old. And we intend to be around for a while!

Feel free to hit me up with questions!

Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by jhdavids8 on Tuesday October 11, 2011 @06:07PM (#3100564)
I contemplated websolr recently but went with IndexTank simply due to the free storage they provided. It's nice to test out the service in dev mode to see if it's a good fit before committing to paying for it. Any chance Websolr offers an option like this down the road?

Plus, Solr just seemed to struggle with geo search. I used Sunspot as my search interface and had to specify a minimum precision of 6 (supposedly a 6+ mile bounding box) to get any results, even those that were less than a mile away. I know Solr uses geohashing, but it just appeared wildly inaccurate...at least for me.

Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nzadrozny on Tuesday October 11, 2011 @06:17PM (#3100613)
> Plus, Solr just seemed to struggle with geo search.

I hear you on that. Solr only relatively recently got a decent, official Spatial search API with its version 3 release. We've got a Solr 3 cluster up at Websolr (with lots of other tasty stuff, like real-time updates), but Sunspot is lagging behind a bit.

I'm one of the maintainers of Sunspot and we're pushing out a new 1.3 maintenance/bugfix release within the next few days, which should clear the way for Solr 3 support -- and much better spatial search -- in Sunspot 1.4.

Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by jhdavids8 on Tuesday October 11, 2011 @06:26PM (#3100654)
Fantastic. I just sent the email. I'll be on the lookout for Sunspot 1.4 and its improvements in geo search. Thanks!
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nzadrozny on Tuesday October 11, 2011 @06:14PM (#3100597)
> Any chance Websolr offers an option like this down the road?

We're definitely open to adding easier free trials. As a developer myself, I respect when a service meets you half-way on that initial integration cost.

In the mean time, we're pretty liberal with trials. Hit me up at [email protected] if you're interested and I'll hook you up with a free month to try things out.

Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nostromo on Tuesday October 11, 2011 @05:58PM (#3100503)
You list Java as supported on this page, but the link 404s. http://websolr.com/guides/solr/clients Do you guys support Java integration?
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by cedrichurst on Tuesday October 11, 2011 @06:07PM (#3100562)
I don't work for websolr, but I do quite a bit of work with Apache Solr. I'd imagine Java integration is supported through the native Java client, http://wiki.apache.org/solr/Solrj, which is maintained by the Solr team. There are also a number of Solr client libraries for Python, Ruby, PHP, etc. All searches against a Solr index take place over HTTP through an RPC interface, so building an API for a specific language relatively straightforward.
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nzadrozny on Tuesday October 11, 2011 @06:14PM (#3100599)
Thanks for the heads up on that 404, I'll take care of that. See cedrichurst's comment for what I ought to be linking to.
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by holdenk on Tuesday October 11, 2011 @08:16PM (#3100996)
I'm kind of curious as to what sort of performance improvements your normally able to get for people?
Re:LinkedIn Buys Real-Time, Hosted Search Startup IndexTank by nzadrozny on Tuesday October 11, 2011 @08:41PM (#3101027)
Performance improvements relative to... what?

We do get a surprising number of people who are upgrading from SQL LIKE queries. In a situation like that, you might see queries that take a few seconds to scan through a table in SQL get cut down to a few dozen milliseconds to fetch results from an index. That's just a good example of using the right tool for the job.

It would be a little tricky to compare performance to self-hosting, given the myriad of variables involved. And as far as other services go, perhaps someone else can speak to that. Or maybe you have a more specific scenario in mind?