One of the great things about the un-conference is the spontaneity of the ideas. In the morning I ran into an old colleague whose startup was looking at switching databases and struggling the options. Hence, “Scalable Databases for Startups” seemed like a great topic, and we were off and running full steam after lunch.
The session brought in a wide variety of startups. While there were several vendors there – Basho, Calpont, InterSystems, ParElastic, and Tokutek (my firm), there was also fortunately a number of startups and bigger companies alike willing to share stories, including Care.com, Curata, Iron Mountain, Lucidel, and Mapkin.
First of all, we found that there was no single solution out there that all startups were either using or gravitating towards. While several folks were using MySQL, SQL Server, and contemplating MongoDB, we also heard about Hadoop, Sphinx, CouchDB, PostgreSQL and the related Google App Engine being considered. Basically, companies were struggling to find the right fit of the menu of options.
|No shortage of options being considered!
A common beef folks had was with the ubiquitous off the shelf MySQL offering. Pawan Deshpande of Curata complained that it wasn’t the best fit for dealing with big BLOBs or unstructured data. Dave Krupinski of Care.com noted that standard MySQL lacks performance at scale, has inflexible schema, and that replication and backup can be a challenge.
So what are folks looking for in alternative solutions? Solutions that proven topped the list, as there are so many new database companies out there. For example, there were concerns from some users on a popular NoSQL solution that is easy to install but has trouble sharding andscaling, especially due to their locking upon writes. Other NoSQL solutions discussed offered more guarantees, but lacked performance.
On a final note, folks are wading into the cloud. One participant noted that they are starting in the cloud with Amazon for new/international offerings to test the waters. The trade-off is having a level of abstraction and giving up control, but in return getting a good service, In fact, they noted that if they could do it all over again, they would actually have putmore of their legacy apps in the cloud.
At the end of the day, despite all the messaging from vendors, personal recommendations and feedback seemed matter the most. The last 10-15 minutes was spent with some serious networking and 1xN discussions oftechnology choices, hopes, and gotchas.
|Nothing like hearing the experience of peers.