Eight Reasons Why Avalanche is 8X Faster Than Snowflake By Emma McGrattan October 28, 2020 This month, GigaOm published a comprehensive evaluation of leading cloud data warehouse services based on performance and cost. Offerings analyzed included Snowflake, Amazon Redshift, Microsoft Azure Synapse, Google BigQuery, and our very own Actian Avalanche. There are many intriguing results included in the report, but an indisputable conclusion was reached by GigaOm: “In a representative set of corporate-complex queries from TPC-H standard, Actian Avalanche consistently outperformed the competition.” To put “outperformed” in more concrete terms, Avalanche was 8.5X faster than Snowflake in a test of 5 concurrent users. In terms of price-performance, the advantage over Snowflake was 6.4X. Avalanche was built for performance out-of-the-box So what’s the secret behind Avalanche’s superior performance? There is a simple, fundamental explanation. Avalanche was built from the ground up to deliver unrivaled performance on commodity infrastructure. Its original design goal was to makes the most of every CPU clock cycle, every byte of memory, and every I/O operation. And ensuring high performance continues to be a priority for us. Our efficient design is the reason why, even with the limitless resources of the cloud, you won’t see your costs ballooning as you increase concurrent users or data volume. How specifically does Avalanche deliver best-in-class analytics performance without the need for tuning? It is a combination of the following eight factors. Vendors such as Snowflake may offer a few of the capabilities listed such as Vector processing, but the unique combination creates a powerful compounding effect: Optimizing the use of microprocessor cores to run multiple data operations in parallel during a single CPU cycle. This is known as Vector processing. Traditional scalar architectures typically consume myriad more CPU cycles to compute the same calculations, which impacts overall throughput. Taking full advantage of multi-core CPUs – Avalanche can perform Vector processing across all available cores, which maximizes concurrency, parallelism, and system resource utilization. Processing data using the CPU’s on-chip cache is faster and closer to where operations are performed and therefore optimizes performance. Our competitors tend to use DRAM for query execution cache, which is far slower. Using advanced compression – typically with a 5:1 compression ratio – Avalanche’s compression algorithms are designed for maximum efficiency, particularly in decompression, yet still deliver about a 5:1 compression ratio. Compression is handled automatically, so no tuning is required. Optimizing I/O – Avalanche is a pure columnar implementation. The data lives its life in columnar format, which results in I/O efficiency. Using patented technology to maintain indexes automatically so that an indexing strategy is not necessary. MPP architecture parallelizes query execution within and across nodes to power through business workloads regardless of size and complexity. Real-time updates that enable operational insights with zero performance penalty enabled by Avalanche’s patented positional delta trees. All this adds up to blazing-fast analytics performance that results in faster iterations on data models, quicker root cause analysis, and ultimately enabling a data-driven organization. When cost is more important than performance If you’re satisfied with the current performance of your data warehouse, Avalanche can deliver that same level of performance while enabling you to dial back your spend on compute resources—which can immediately translate into considerable cost savings. Conversely, if you can benefit from increased performance, Avalanche can give you much more at a much lower cost. In other words, you have two levers to play with – price and performance – and Avalanche enables you to achieve the lowest cost of ownership for the level of performance your business demands. Cut your Snowflake Bill in Half – Guaranteed Actian is currently running a guaranteed offer to cut your Snowflake bill in half. The offer is not based on an artificial discount but rather on our confidence that our technology will deliver superior price-performance over Snowflake, as proven by GigaOm’s authoritative report. Want to Learn More? We encourage you to watch the GigaOm webinar featuring the report’s lead author to understand the study’s methodology and the implications of the results in greater detail. To learn more about the innovative architecture that powers Avalanche’s high performance, check out this video from our SVP of Engineering, Emma McGrattan. About Emma McGrattan As SVP of Engineering at Actian, Emma leads research and development for the Actian Vector, Actian Vector in Hadoop, Actian X and Ingres teams. A recognized authority in DBMS and Big Data technologies, Emma is a sought-after speaker at industry conferences. Emma has recently celebrated over 25 years in Ingres and Actian Engineering. Educated in Ireland, Emma holds a Bachelors of Electrical Engineering degree from Dublin City University.