Dive Brief:
- Google Cloud upgraded the performance of its Cloud Spanner database service in a direct pricing challenge to Amazon's DynamoDB, according to a Wednesday blog post.
- The improvements will effectively let customers run Cloud Spanner at “half the cost of Amazon DynamoDB for most workloads,” the company said.
- While pricing won’t change, performance improvements will increase transaction speed by 50% and more than double storage per server, from 4TB per node to 10TB, according to Google.
Dive Insight:
In pay-as-you-go cloud ecosystems, decisions about how and where to house vast stores of rapidly expanding data sets can have an outsized impact on IT’s bottom line.
The emergence of data-hungry LLM technologies has only raised the stakes, as organizations look to leverage AI-enabled tools.
The three largest hyperscalers — AWS, Google Cloud and Microsoft — are in a heated race to optimize data centers for AI workloads, integrating advanced GPUs and other high-speed processors to their infrastructure.
Last month AWS invested $5 billion in AI company Anthropic, which will build its models using the cloud giant's AI-optimized Trainium and Inferentia chips. Google Cloud unveiled its latest generation of TPUs for model training and inference workloads in August. .
Processing speed is a key variable, both for efficiency and optimization.
The announcement pointed to an Amazon blog post touting DynamoDB's performance during Prime Day, when it processed 126 million requests per second. AWS’ managed database solution powers Amazon’s websites, its virtual assistant Alexa and the company's network of fulfillment centers.
Cloud Spanner can handle three billion queries per second, according to Google.
The price-performance improvements are currently available in select Google Cloud regional and multi-region instance configurations, with all other configurations to follow, the company said. Storage upgrades will roll out in the coming months.