So, the Heroku plugin goes away and customers like Reddit have to make other arrangements. Interesting that they would shut down a viable business - I don't understand that.
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!
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.
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.
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 firstname.lastname@example.org if you're interested and I'll hook you up with a free month to try things out.
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?
Super not happy to find a new search solution in the next six months. :(
I guess that's the new way to run a business. Exit > customers.
Ideally we'll get other providers to implement the IndexTank API for a seamless transition. I will post updates about this.
We use Flying Sphinx and it's worked out well.
But, I wish they held off on the acquisition until other "cloud providers" were willing and ready to implement these services that will be open sourced. Or at least keep your service up until then. This is great for the IndexTank team, but screws all of their customers (of which I am one).
Very intrigued to know which specific components of their technology will they open source - and how they fit within linkedin sna different projects http://sna-projects.com/sna/ and lucene / other open source projects at large.
Really curious to know if any IndexTank / linkedin team member reading.
That said, it proves once again that entrepreneurs should not feel discouraged by the presence of competitors in their field. Continue to build your own great product. Your biggest, most successful competitor may be the next one to fail for whatever reason or, hopefully for them, have an early exit which clears the field for your product/service.
Congrats to Diego!
When's the party? :-)