Yr. Obt. Svt.’s Take
OneStream XF 5.2 (henceforth plain old 5.2) is both awesome and odd.
Let’s get the odd out of the way as it’s just my opinion: this should have been version 6.0, not 5.2; there’s an awful lot of stuff here, much of it a huge extension of the product. Why something with roughly five new and frankly huge features isn’t worthy of a more exciting description than a dot release is beyond this geek’s ken and is a mystery best answered by inscrutable marketers and product managers.
Whatever the reason – possibly because it’s the product of Midwestern modesty – what 5.2 brings to the table is just enormous. Here are the highlights:
- BI Blend
- BI Viewer and Dashboards
- Pivot Grid (Standard and Large)
- Table Views
- REST API
We’re talking new UI components (big ones including a new dashboarding feature), new engines (Merciful Creator, again, how can this be a dot release?), new connections, a new API, and a bunch of other features. This is a big, big, big release. This blog post can only cover the highlights – subsequent posts will try to review each feature in turn.
With that, let’s begin looking at these tools.
I don’t want to harp on this (totes obvs I do as I’m mentioning it), but an aggregate storage (I believe columnar so 21st century stuff) model for large data sets all by itself seems to merit more than adding a 10th to a release.
Beyond my whining, this is Big Stuff. Want to scale far beyond the “normal” OneStream engine? Done. Want to support large data sets and dimensions? Done. Want to point to transactional data stores with rapidly changing data? Done. Want to in essence have little to no difference in the UI? Done. You’d be an odd duck if you didn’t want all of this. The only thing, and I mean the only thing I’d like to see is making this read/write instead of read-only. Consider this an unofficial enhancement request bereft of understanding why it is the way it is. I shall struggle manfully and gleefully onwards with what again is a huge leap forward in the tool.
BI Viewer and Dashboards
I saw the drag and drop fancy dashboarding (really, it was and is) that incorporates data visualization at Splash this year and the cynic in me thought, “If it works half as well as it demos, it’s a game changer”. I’ve not kicked the tires but am very much looking forward to it. Given that I have the artistic ability of a very untalented two-year-old, any help I can have with making graphics pretty is most welcome indeed. Did I mention that it can go against any number of data adapters? Interesting stuff.
Aka pivot tables – yep, real user-driven custom reporting in the tool. So cool – I guess at heart I’ll always be fascinated by data.
A pivot grid with all the trimmings – a panel/palette for dimensions, drag and drops into a grid, aggregated data, conditional formatting, custom calculated fields, sorting, grouping, etc., etc., etc. along with drill down to the source data adaptor.
The standard edition pivot grid pulls the data into local client memory. This is the major differentiator between standard and large as sooner or later, a client will run out of memory.
Roughly 100,000 rows is a reasonable limit. I’d guess (see below) that this will be influenced by memory, speed, network, etc.
If the data is processed in slices and does that on the application server (which reasonably will have more available memory and significantly faster CPUs), larger data sets can be addressed. Unsurprisingly, this means that, depending on server speed, load, proximity to data, and efficiency of the data source, performance may be impacted. OTOH, why not just go ahead and outfit your company with $14,304.01 Dell laptops configured thusly: Xeon E2286M, 8 cores, 16M cache, 2.4 to 5 Ghz (turbo), 45w, vPro, Nvidia Quadro RTX 5000 with 16GB GDDR6, 17.3” Ultrasharp display, 8 TB of SSD disk laptop with 128 GB of RAM. I reckon it’ll probably handle those larger data sets without using the larger data grid. Yes, I really did price that out and no, it probably isn’t a terribly realistic solution. Perhaps it is best to follow OneStream’s lead?
Per the docs, an external data source (it appears that the application database itself can be the source for these grids as well as the BI Blend engine tables as well), is anything a database adapter can point to. So far tests with two to four million have worked. Remember, this is a pivot table so that’s rather a lot of data to directly analyze.
One thing to note – pivot tables of either source are not OneStream applications in themselves (this is sort of the point as otherwise why would you use anything other than a Cube View or Excel or a report or whatever) so all of that lovely financial intelligence you’re used to won’t be there. Understand your data – I imagine there will be a brisk uptick in looking at financial tags in source data and flipping signs to get aggregations to report correctly.
A few other notes:
- There are four types of aggregations: sum, average, min, and max
- The pivot table (again, either source) resides within a custom dashboard.
- Groupings and aggregations occur in the tables (think queries), not in the grid itself. This isn’t Excel’s native pivot tables but something more like Power BI.
- There’s all kinds of lovely conditional formatting
- Calculations can be applied against members. I have no idea how this works (eh, you wouldn’t be reading a blog by Yr. Obt. Svt. if he either had all the answers or didn’t tease you with hints of future blog posts or induce you to possibly vainly hope that he’ll ever get any of it truly right) but it looks absolutely tippy-top at first blush. Seriously, there’s an Expression Editor module that suggests all kinds of lovely calculations, I’m too lazy to have installed 5.2 (yet another future blog post), and I’m not going to speculate beyond that. Stay tuned.
OMG, reading through the docs is dizzying. There is a lot of functionality. A lot. And it isn’t clear to me if it’s in both grid types or just one or…. I’m going to stop trying to summarize this in the all-too-likely likelihood that I’m going to miss something really important. More to come and I’ll have real details.
Some other functions, both minor and major
It’s sort of beyond me at this point to identify what is major and what isn’t as there’s so much here. I’ll do my best to give you the most obvious ones and you’ll have to categorize their importance.
I’m pretty sure this one is major – data automation is now addressable through a JSON format.
Table Views for Spreadsheets
A way to embed disparate data sources into application spreadsheets (so within OneStream, not Excel) including relational data.
Spell Checker for Data Cells
This one really is minor but perhaps not — be glad that only Microsoft Word sees the horrors of my spelling. Now OneStream can share the wealth.
I’m really and truly going to stop writing about this release and start installing, building, and using 5.2.
I’m also going to really and truly try to explain more using less words. This
will absolutely never probably won’t happen but it’s still a laudable goal.
Be seeing you.