Collage.com is a high-transactional ecommerce site that allows anyone in the world to create meaningful and memorable gifts out of photos. They ingest and analyze hundreds of millions of rows of data every month, which enables the business intelligence (BI) team to generate key insights and make better decisions. They use MySQL as their production database and replicate it to Amazon Redshift.
Nick Smith, senior software architect at Collage.com, knows how important data replication and accessibility is for his company, but it's not something he wants to spend time worrying about.
"I have a lot of other stuff going on. Reviewing technical requirements, operations work, and a ton of hands-on coding," said Smith.
So when Collage.com's previous ETL-as-a-Service provider stopped supporting Redshift, Smith's team prioritized factors like ease-of-use, reliability, and technical support when searching for an alternative.
FlyData vs. Fivetran: Comparing ETL-as-a-Service Vendors
Smith and his colleagues reviewed every major ETL-as-a-Service vendor, including Stitch, Blendo, and Xplenty. After an extensive search, they narrowed their options to FlyData and Fivetran.
In the end, Collage.com chose FlyData over Fivetran for several reasons:
- Technical support
- Table structure
- Replication cleanliness
- Configurability and frequency of replication
- Migration costs
- Straightforward pricing
Smith has been surprised and delighted by the response time and technical competence of FlyData's support team. "FlyData's level of support is far above most other 3rd party vendors we work with. It's been phenomenal," said Smith.
Smith's team wanted table structures in Redshift to match as perfectly as possible to those in mySQL. That way, they could write queries for both without thinking about it. "A query we write for mySQL should work in Redshift, and vice versa," said Smith.
FlyData replicates the table structure perfectly from mySQL to Redshift. Fivetran did not. "Fivetran made you create a view for each table in Redshift, which was a no-go," said Smith. "We can't manage that many views."
The Collage.com team needs their replicated data to be identical with their production data. FlyData has out-of-box functionality to ensure this, including automatic schema change replication.
"The cleanliness that FlyData replicates from mySQL to Redshift is really important to me," said Smith.
Unfortunately, Smith found serious problems with Fivetran's replication cleanliness. "Using Fivetran, the data that gets deleted in MySQL does NOT get deleted in Redshift — it just gets 'marked' as deleted. That was a huge issue for us," said Smith.
Configurability and Frequency of Replication
Collage.com's BI team needs the most up-to-date data at a moment's notice. Smith's team found that Fivetran synced in roughly 5-minute intervals. FlyData, however, syncs in near-real time.
"FlyData won this category," said Smith.
With FlyData's schema structure and cleanliness of replication, migration was a breeze for Collage.com. "It's very transparent. The cost of migration to FlyData was basically zero," said Smith. "With Fivetran, we were going to have to do a bunch of work."
Smith was hesitant to lock themselves into Fivetran's one-year contract requirement. On the other hand, FlyData offered month-to-month subscriptions that were competitive with Fivetran's annual pricing.
"FlyData won this one, too," said Smith.
FlyData: Worry-free replication to Redshift
The holidays are crunch time for Collage.com. They need flawless execution across the organization, and particularly from their data infrastructure.
When failure's not an option, FlyData provides Collage.com with dedicated bandwidth to ensure they can sustain peak volume times. It's just one more reason why Smith and his team love FlyData.