AWS Have Big Plans for the BI Industry
- 22nd June 2018
- Peak Indicators
The AWS Summit was held during May, with 3 of the Peak Indicators team in attendance. Our Services Director David - who was at the event with Shan and Joe - has shared his thoughts from the Excel Centre:
At Peak Indicators, we are speaking to lots of Birst customers who are heading towards using Azure or AWS. As a company, this means that we are re-evaluating the direction our clients are taking; whilst also seeing how these alternative technologies stack up against Oracle Cloud.
The scale of AWS is simply mind-boggling. They are a 22 billion dollar company in terms of revenue already - but they are now growing at a rate of 49% a year. Even more incredibly, that is a figure that is accelerating; since growth was 42% two quarters ago.
Even in terms of the AWS Summit alone, the event has grown from 3,000 to 10,000 in just two years.
Interestingly, the fastest-growing product ever is Amazon Aurora - a MySQL/PostgresSQL database. This software:
- Costs $0.10 a month per GB
- Costs $0.20 per million requests
- Is automatically backed up to 6 data centres, across multiple zones
- Is 1/10th the cost of an alternative, on-premise solution
A key focus for this year’s Summit was Machine Learning, which remains a hot topic in the world of business intelligence. Whilst Amazon discussed ideas of their own, the area ties in with our very own machine learning service - Tallinn - which we launched earlier this year.
We also heard time-and-time again about Amazon Lambda - an event-driven, serverless computing platform provided by Amazon as part of Amazon Web Services.
Why? Well, everything Amazon are doing is moving towards serverless computing. For example no VMs, with just containers to orchestrate services.
What About Analytics?
In terms of Analytics, we spent some time with one of the Amazon Architects to understand what they provide. Looking at the AWS offerings, you can create similar systems to OBI and Birst Stack with the following:
- AWS Glue is effectively the physical connection and ETL piece
- AWS Athena is the SQL-based query engine (think logical layer in the rpd)
- AWS S3 is the storage (or AWS Redshift for very large datasets)
- AWS Quick Sight is a very cheap and simple front end
It’s worth noting these aren't installed and they aren't VMs - they are just services that we call.
So the takeaway:
The scale Amazon are operating on is just vast. It will be interesting to see how rivals approach the challenge of competing with them.
We spoke to a Birst staffer who has joined Snowflake, a high-end, Amazon-based database. The clever thing with Snowflake is that Amazon separates out the compute element (very expensive) and the storage (very cheap).
So what does this mean? Well, you only fire up the compute power when you need to reload your data. For example, one North American customer who fires up 1536 nodes (servers) once a month, for a couple of hours, then uses 0 nodes until the next load.
To buy that kit would cost millions, whereas those using Snowflake only pay for what they use. As a result, Amazon are currently signing up 15 customers a day to the software!
We’re excited to see how AWS will shape the future of analytics. From what we saw, the effect it will have is set to be significant. We’re intrigued to see how Azure, Google and Oracle respond.
Managed Analytics Services
Interested in learning more about what managed analytics could do for your company? Peak Indicators can help. Our analytics services allow you to focus on your core business priorities whilst improving the overall performance of your business.
Contact our team of BI experts today to start your own analytics journey.