Pinecone’s Latest Enhancements Usher in Savings Opportunities with Improved Vector Search Efficiency

Pinecone, innovator of vector databases for building high-performance vector search applications, is debuting its latest enhancements to vector search that address ongoing challenges for the system. Issues such as determining type and size of index needed for data and performance needs, supporting high throughput, and scaling up indexes without interruption are no longer pain points for Pinecone users. These improvements take shape in features like vertical scaling, collections storage, and p2 pods for optimized vector search usage.

Vertical scaling introduces scaling index capacity with zero downtime, enabling pod capacities to be doubled for a live index without time wasted, according to the vendor. No longer will users have to create new indexes with more pods to accommodate index capacity peaks, eliminating the expenditures of valuable resources like cost and engineering time.

“Vertical scaling means no more migrating to bigger indexes or writing to an index already at storage capacity,” said Isabella Fulford, software engineer at Mem Labs. “This is going to be a huge timesaver for us.”

Pinecone’s collections feature provides users with the ability to save data from an index as a snapshot, and then create new indexes from any collection. This function creates a variety of potential use cases, such as utilizing collections for backing up and restoring indexes, testing different index types with the same data, and migrating data to a new index. Future updates will see collections allowing for imports and exports of data to and from S3 or GCS blob storage, as well as write streaming and bulk data directly to collections, according to the vendor. Storage costs for collections will be $0.025/GB per month for all Standard, Enterprise, and Enterprise Dedicated users. Users on the free Starter plan will be able to have one collection at no cost.

Introduction of the new p2 pod type accommodates users with high-throughput applications, like social media apps or streaming services, by implementing accelerated search speeds under 10ms and throughput up to 200 queries per second per replica. This provides up to 10x lower latencies and high throughput than the p1 pod type, according to the vendor. The p2 pod type consists of a new, graph-based index that trades off ingestion speed, filter performance, and recall in exchange for higher throughput; it still supports filtering, live index updates, and all other index options, according to the vendor.

Other improvements to Pinecone include faster indexes on s1 and p1 pods, addition of s1 pods to their Starter plan, as well as multiple areas of cost-saving opportunities.

Updated pricing for p1 and s1 pods will go into effect for all new users as of September 1, 2022, starting at $0.096/hour and up depending on plan, pod size, and cloud environment, according to the vendor. Existing users with a running index on a paid plan as of August 31, 2022 will not be affected by the price update, and will maintain their current rates for s1 and p1.

For information about this update, please visit