GraphScope
GraphScope copied to clipboard
alibaba
🔨 🍇 💻 🚀 GraphScope: A One-Stop Large-Scale Graph Computing System from Alibaba | 一站式图计算系统
- [x] load single partition of Sf-30k graph on ecs, evaluate graph loading time and memory cost - [ ] load Sf-30k graph on ecs, run simple bi query (bi1,.etc)...
## What do these changes do? Add a prometheus exporter for coordinator to monitor computing performance.  ## Related issue number [#288 Monitoring for GraphScope](https://github.com/alibaba/GraphScope/issues/288) Fixes
This includes a series bug fixes and enhancement of groot deployment.
Signed-off-by: bifenglin ## What do these changes do? change `session.py` `dag.py` `operation.py` and `context.py` to persistence and restore the data ## Related issue number #287 Fixes
- [x] Profiling performance of fragment-based RDD and. - [x] Profiling graph computing framework. https://yuque.antfin-inc.com/7br/trqsp0/hr7co5
- Make the system modularized, each component can provide services, restart, and horizontal scale independently. This includes the GAE (could be further divided into GAE-base, GAE-Java, GAE-GPU), GIE (Groot, GAIA,...
We noticed that the quering service is abnormal after pod failure and restart. - [x] Support failover for executor - [ ] Add CI tests for pegasus failover - [...
**Is your feature request related to a problem? Please describe.** Optimize structure of `Record` in GAIA-Runtime. Now `Record` is a little bit complex and may involve extra cost.
## What do these changes do? Add a method `LoadFragmentWithLocalVertexMap` to support load arrow fragments with `ArrowLocalVertexMap` The `local_vertex_map` currently only support in `LoadFragment` with whole batch vertices and edges,...
