The latest Graphistry release makes GPU-accelerated visual graph analysis easier for data scientists and teams and grows where you can use it. The current release hits many themes of our mission of bringing 100X investigations through GPU visual graph intelligence to all analysts:
- Affordability – Azure: Azure: Experiment with T4 GPUs for less than the price of coffee
- Collaboration – Sharing: Securely collaborate widely with the new sharing panel and APIs
- Data science – Easier team data science & research with JupyterLabs preconfigured for GPUs and more
- Ecosystem – Gremlin Python connectors that make Graphistry the most native tool for exploring Neptune, Cosmos, and other Gremlin graph databases
Read on for these and more, and you can always check the full release notes. Many of these features — and major internal version & security upgrades across all our JavaScript components — have been getting great daily use on Graphistry Hub, so we’re happy to have them hardened and brought together.
Also, if you haven’t heard — we’re hiring and would love chat!
Azure: Experiment with T4 GPUs for less than the price of coffee
Graphistry is on a mission to make GPU visual graph intelligence easy and accessible for most people needing to analyze data, so we consistently work to drive down costs and pass the savings to our community. Excitingly, we just took a big step for our Azure community through the new T4 GPU instance types.
For much less than the price of coffee, you can now launch private Graphistry enterprise GPU experiences on Azure with the affordable Standard_NC4as_T4_v3 instance types. The T4 option does not yet appear on the pricing page: go to Get it now -> Create -> Instance Details: Size: See all sizes / Request quota.
Figure: New sharing panel
Scale secure collaborations with the Sharing Panel and API
import graphistry
graphistry.privacy(mode='private')Figure: API mode for new sharing capabilities
Collaboration helps analysts scale their reach and impact, so we are excited to launch the sharing panel. When ready, use the share button to invite your colleagues, and as needed, lock down who can do what. Access is initially for Hub Pro users and all self-hosted + marketplace ones. We especially love that sharing controls are easy to use both from the UI and, via the API’s one-liners, Jupyter notebooks.
For more information, we strongly encourage checking out the deep dive in our recent post “The sharing paradox: To scale collaborative investigations, Graphistry has been locking things down.” Over the next few months, we’ll be releasing successive waves of features around security and sharing. Much of the current release was around carefully adding access control infrastructure and content tracking for enabling these kinds of features, so we are excited.
Team GPU data science with JupyterLabs
Figure: JupyterLabs with GPU monitoring, file browsing, and more
As heavy Jupyter users, we’ve been periodically polling our community for if/when the majority want Graphistry’s integrated Jupyter environment upgraded from Notebooks to Labs: it’s finally happened! Launch Jupyter from the menu link in your Graphistry instance as before, and now, get an updated Labs-mode experience.
JupyterLabs comes out-of-the-box with great features, and we’ve been adding integrations specific to our community:
- Built-in file browser & search
- RAPIDS.ai-ready kernel (as before)
- dask-cudf runtime exposed for scaling notebooks & monitors
- Live GPU resource activity monitoring UI
If you have additional extensions and configurations you would like built in, we’re listening.
TinkerPop Gremlin: Nexus, Cosmos, and other Gremlin databases
import graphistry
g = graphistry.neptune(endpoint=my_endpoint)
#g = graphistry.cosmos(COSMOS_ACCOUNT='', COSMOS_DB='', COSMOS_CONTAINER='', COSMOS_PRIMARY_KEY='')
g2 = g.gremlin('g.E().limit(10000)').fetch_nodes()
print('nodes dataframe (# nodes, # node attributes):', g2._nodes.shape())
g2.plot()
Figure: PyGraphistry Gremlin integration for combining Amazon Neptune / Microsoft CosmosDB graph database, pandas dataframes, and Graphistry GPU graph visual analytics
We have been investing in bringing GPU-accelerated visual graph intelligence to the Gremlin ecosystem, starting with improving the Python bindings. Our latest PyGraphistry API update makes it easy for data scientists and developers to wrangle Gremlin-based graph databases with PyData ecosystem:
- Query Gremlin databases and get results as dataframes (pandas, arrow, …)
- Upload small property graphs as dataframes
- Explore/embed the results as GPU-accelerative interactive Graphistry graph visualizations
… and more!
The release comes with additional features, fixes, speedups, and upgrades, many of which are in the full release notes. Happy graphing, and swing by our Slack channel for ideas and help. Likewise, if you’re thinking about a career move, we’ve opened several roles and would love to chat!