5 ways to reduce advertising network latency

Instead of having to upload stuff into the Google cloud and then feed it through to BigQuery, cloud wranglers can now stream data directly into the Chocolate Factory's analytics-as-a-service tech.

Google announced the "streaming data" update to BigQuery on Wednesday and it means admins can pour data into the service and start analysing it quickly, rather than waiting to upload it into the Google cloud storage layer in bulk.

Doing this speeds access to data and real-time insertions, Google says, but risks "possible data duplication" and "possible connection drops."

Appropriate applications for the service include ones that are constantly generating data that needs need quick-turnaround analysis, such as dashboards.

"The new data is available for querying instantaneously. This feature is great for time sensitive use cases like log analysis and alerts generation," Felipe Hoffa, a Google developer programs engineer, said.

However, limits apply for streaming: the system can ingest 100 rows per second, per table, though Google will permit "occasional bursts of up to 1,000 rows per second." The maximum row size is 100KB and the maximum data size per streaming insert is 1MB.

The upgrade follows Google introducing SQL-like commands such as JOIN into the technology in March, and cutting the price of data storage costs from $0.12 per gigabyte to $0.08 in June.

Streaming data will be free until January 1, 2014, at which point it will cost 1 cent per 10,000 rows inserted, Google wrote. ®

5 ways to reduce advertising network latency