Feedly is fenomenal, its my 2nd most used app after email. I really love the advanced features of mute filters, web alerts and Leo - Feedly's AI Engine.
The issue started occurring intermittently at 08:26 US/Pacific. The issue is impacting Google’s Backbone network and may impact various services when accessing them from a different region or from the internet. Impacted services include Cloud Services (Workspace, Firebase, GCP) as well as other Google properties.
Connectivity within a zone should not be impacted.
Our engineering team has implemented a mitigation and is now monitoring the effectiveness of the change.
BigQuery with it's $20/TB for active, and half of the price for cold storage, is the best structured (eg: columns or JSON) solution where import and export is free. You even get SQL as the product is a data warehouse.
Google Cloud Platform - their PaaS or serverless solutions are extremely cheap. Also having 1TB free query on Bigquery is a true hidden gem in cloud. The same instance compared to other cloud vendors have better performance. We always needed fewer workers on GCP compared to AWS for example.
As 1TB is free for processing every month, using SQL 2011 standard + combined with Javascript UDFs, the winner solution is Google BigQuery for us, combined with Dataprep
During the 4 years we at REEA.net tried a lot of approaches, in different projects.
1. We stream every upate as a new row directly to the table. We have for every table a view that has a naming convention "table_latest" and it targets latest version of the record. We didn't see a big latency increase.
Deleted rows have a 'deleted' flag, we use EXCEPT in the query.
2. Since DML is available we started on newer projects dual aproach, having a dedicated streaming table, and rematerializing to a flat table. It adds developer complexity but it's a win, as the materialized table is easier to be used by our marketing team.
We added to the reports "Based on data 10 minutes ago" label, and a CTA "re-run using live data", which in background calls the script to rematerialize the table and rerun the report. Our markting team loves this, and they don't abuse this functionality. On some projcts we have remateralization every 1h, on others daily.
3. regarding partitioned tables where you can stream only 7 days of window into their respective partitions we use also a complementary load job. We stream in the 7 day window. And we have a heavy "future date" use case, where we use "load jobs" to a temp table, then query and write to their respective partitions.
We are using BQ for 4 years now. Since the beginning with dual-writing to diferent engines. We did an architecture that routes data to Redis, Riak, Elasticsearch, Mysql, BigQuery.
We often have 500+ lines of queries in BigQuery writen in standard SQL that runs our reporting system. The average query time we have is 10 seconds usually scanning at least a terabyte.
We truly love the ability to write in Javascript the UDFs, and our marketing team loves to combine their Google Sheets (mapping data) with a real BQ query.
We heavily use the 'dryrun" feature of BQ to see without running the query how much resources and costs will generate. We even implemented a quota for users, based on this feature, wich works great, and only when they do a mistake in a query targetting alerts them at a certain treshold.
Afer a couple of years we naturally started to consolidate all our data into BigQuery.
Since it's Standard SQL all our team members can start using and reporting + tooling works quite well. DataStudio is another service which we started to use more and more as well.
It's been a golden year for games of many genres. For example some amazing shooters from 2016: Battlefield 1, Titanfall 2, COD: Infinite Warfare. Even the worst of them is still a piece of art compared to the best shooter of a few years back.