posted: March 8, 2017 | author: Marianne Cardwell
This is a big year for Woolpert in terms of GIS: we have been Esri Business Partners for 20 years! Below is a picture of Layton Hobbs, Woolpert Director of IT and R&D, myself, and Jack Dangermond with a plaque for our 20 years of partnership.
Woolpert sent a big team to both the Partner Conference and Developer Summit this year. Three of us attended the Partner Conference and five of us are attending the DevSummit! As always, some statistics:
This is Esri’s 12th DevSummit and its format is similar to past years. I don’t remember sessions going until 6:30PM last year, but my memory may be failing me! These are long days, but we’re all learning valuable information.
Today’s big focus was the plenary, which always gives us a very good overview of what’s to come. Below is an overview of what stood out to me. Esri live-streamed it and the video is already available on their site. If you have just a little bit of time to get the big picture, I highly recommend watching the Plenary. Esri’s done a great job of dividing it up into segments so you can easily watch only what’s important to you.
There were a few new languages or APIs that were specifically discussed today. I’m pretty sure we’ll use at least some of them:
ArcGIS Online continues to be a big focus, of course. Here are some ArcGIS Online statistics, which I thought were quite interesting:
Esri makes regular improvements to their SaaS platform. They highlighted improvements in their CDN for AGOL. Their CDN is available globally and is not only applicable to the API but also to their data, such as raster and vector tiles and scene services.
Esri’s on-premise offering of ArcGIS Online, Portal for ArcGIS, is of course taking on a bigger and bigger role. ArcGIS Enterprise is composed of the GIS Server (your typical ArcGIS Server installation), Portal, data store(s) and the web adaptor. More and more functions require the use of Portal or ArcGIS Online. One new offering here is the concept of Portal to Portal collaboration. This may be particularly useful in organizations that may have multiple Portal instances that need to talk to each other and share information. You still keep both separate, but you can now share specific data, maps, and applications between the two, allowing for better collaboration.
One area that I find particularly interesting is data science. This is a growing field and is clearly a growing area of focus for Esri. This is best exemplified with their new Insights and GeoAnalytics Server products. These both were released in December. Insights is a web-based interface that includes drag and drop functionality while the GeoAnalytics Server tools can be used from both the web and Pro. Both products are still very young and Esri is interested in seeing how we are using them and what we are looking for in terms of functionality moving forward.
Still related to data science was the discussion of the R ArcGIS Bridge, which allows us to use R for analyzing ArcGIS data. The library provides functions to load ArcGIS data. This functionality can then be wrapped into geoprocessing tools that can reference R scripts.
One thing that became very clear to me is the focus on distributed computing. While it is clear that this is a focus on the server level with ArcGIS Enterprise, it is also a focus on the analysis/workflow side with many tools taking advantage of a distributed architecture. For example, when performing raster analysis, while it can be done on the fly based on the current map extent/scale, it can also be run on very large datasets on multiple nodes. This could result in some drastic efficiency improvements.
There were many other things mentioned during the plenary, and I’ll list some of them that stood out to me in no particular order:
I’ve enjoyed most of the keynotes at the DevSummit, but this one was especially well done. Todd Greene, Founder and CEO of PubNub did a really great job. It was informative, funny, and personal. I highly recommend watching the video once Esri posts it online.
Big Data and Analytics
I’ve found it’s often a good idea to come here and determine ahead of time what is important to you. My personal interest is on big data and analytics and so I focused those sessions. I have to say that I am very pleased with what I saw. At 10.5, this is really starting to be possible and viable by leveraging the GeoEvent Server, the spatiotemporal big data store (BDS) and the GeoAnalytics Server. This suite of software, along with a real push for distributed computing, make for a very nice solution. I highly recommend watching some of the relevant presentations once they are posted online. There were sessions that had great visualizations but also lots of sessions on architecting your infrastructure, always with a mention on licensing. These were very helpful in better understanding how all of these components truly work together. Below are a few pictures of slides that I found particularly useful. I’ve read a lot of the documentation online but these really put it together for me.
Another thing that stood out to me was the number of options available for binning. For example, you can perform point aggregation in the GeoAnalytics Server into pre-determined polygons (like state boundaries), or square or hexagonal polygons. With the latter, Esri has a “flat” and “pointy” option, which basically changes the rotation of the polygon: if flat, the top part of the polygon is flat, if pointy, the top part is the vertex.
It’s important to note that there are no additional licensing costs for the BDS. You can have multiple nodes, so you can distribute your data onto as many nodes as you would like, setting up the necessary replication factor.
The GeoEvent and GeoAnalytics servers are both CPU & RAM intensive while the BDS is IO/network/RAM intensive, so keep that in mind when you’re architecting your environment.