This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
Use Case Info: I am essentially using Alteryx as a download / prep / analysis engine to feed data into a universal format client facing software. My workflows are designed around user friendly excel control modules which are accessible on the cloud by the appropriate employees. The current approach allows me to build much more user friendly and flexible formats for our control settings than seems possible via APP construction. The desktop computer developer + scheduler approach has the benefits of limiting the Alteryx licensing costs / training, keeping all management level decisions & adjustments outside of my workflows, and keeping settings in simple protected excel files which allows me to hire non-tech advisers for that role. On its face, it certainly seems to me that a few Designer / Scheduler Licenses is actually a more powerful platform for the cost in my situation. Server appears to me to be geared toward large companies that use Alteryx to generate the final reports / dashboards and not just as an analysis engine. Is that right? Is Server just not designed for me, or am I just missing really important strengths and capabilities that would be of great benefit for my use case? The entry price for Server is around 66k for a Server + 2 Designer Licenses. Plus I will be likely to need additional cores, Alteryx Connect and Alteryx Promote to handle my workload, workflow and workbook requirements. That means it is more likely that I am in at least the 110k range before server fees or inquiring about Promote pricing... If I go with an onsite server, then does this cost just multiply every time I need an additional server? vs The desktop approach by comparison prices in at around 31k for 4 Designer Licenses (2 with Scheduler). That seems like a ton more processing power, flexibility, workflow design capable stations and redundancy for system failures if I go the desktop route. I can certainly buy a few extra rigged out i9 desktops for the price of a the top of the line server. Wouldn’t 3 or 4 fast desktops outperform a single high end server? Until the App control layout can be made as flexible and user friendly as my excel docs, I wouldn't be using the gallery apps in Server to manage my workflow settings either... The only thing that I can think of is that maybe Server might give me the capability to externally trigger a workflow run? That could be a benefit for lower staff workload / manual workflow runs for sure. What else am I missing? Has anyone gone from a multi-desktop Scheduler approach to Server or vice versa? I feel like I have read everything there is about Server and I still don't truly understand the point where the economy of scale starts to tip in its favor...
Kicking off workflows as dependent processes via API calls, version control of workflows in a centrally managed location, user based permissions for who has access to what workflows, analytic apps as well as non apps that can be run on demand by non Alteryx users, dev to prod QA process, etc.
It sounds like you already have your mind made up and your infrastructure may be sufficient for what you are trying to do, but having a centralized server is generally the advisable approach when you want to move away from dependency on the end user machine.
This becomes especially important when you have multiple users needing to collaborate and pass their work to other coworkers. A standardized workflow management system keeps processes out of silos. Just some food for thought. I had access to an Alteryx server in my past life as an analyst and it made coverage of a team members work much easier to manage.
Thanks for taking the time to post some answers. My previous post came across a bit too much like my mind was made up, but I am definitely trying to do an even handed SWAT on whether I should be budgeting for Server / hardware in the next year or 2. I have not messed with an onsite server for over a decade, so I am in the dark performance wise. My main challenge on the Server software side is that the Alteryx pitch seems to be geared toward the benefits of a single large organization easily sharing ad-hoc workflows / outputs. My model on the other hand will have massive workflows, many different clients, almost none of my employees should ever have workflow access, and our organization would do little to no ad-hoc report generating or analysis at the Alteryx level. Combine that with my lack of understanding of Server hardware capability vs desktop, and I feel pretty in the dark
Kicking off workflows via API calls could be a big benefit in data update automation down the line. Thank you for verifying this. Big Plus!
Version control of workflows from a centrally managed location could be a big game changer if I were to take my "client specific" prep containers and split them into separate chained workflows. That would allow me to manage the universal portion of workflows via versions. That had definitely not occurred to me until now and it would dramatically simplify workflow upgrades. Another Big Plus!
User based permissions on workflows is currently achieved through file sharing permissions (and almost no one has workflow or app permission). I am also a bit concerned that giving users the ability to run apps / workflows on server is a serious resource liability. My priority is the speed and stability of the downstream client facing interactive software.
This brings me back to the server vs desktop system resource question. This has become my primary concern of mine as workflows become more extensive. My Alteryx workflows are all processor heavy as they lean heavy on "AI" and incorporate modeling and big data. (I personally hate the term AI as a Sci-Fi fan, but it has become pretty standard...)
Top of the line maxed out desktops are comparatively cheap, fast and designed to multi-task. Reasonably priced servers seem to have slower processing speed, less multi-tasking and are instead designed to organize / prioritize a workload bottleneck. Faster servers that are designed for AI are seriously expensive. I keep reading on the forum that people push their Developer workflows over to Server and may see a massive increase in workflow run times. (sometimes reports of 10x longer or more) I keep worrying about taxing a server on processing workflows, creating bottlenecks for other clients and possibly running the risk of delaying real time data access for client facing software. If I update workflow runs offline with scheduler and just upload static files to a server then I am definitely in known / safe territory speed wise on both ends. I am clueless as far as performance on the server side... "Scaling" is supposed to be a strength of Server over Designer / Scheduler, but this only seems to be discussed on the software side. My server demand / load will always be increasing. What about the cost side of scaling server software & high end hardware to handle an ever increasing resource load? A wrong investment in the server direction for a company our size would be catastrophic. It is hard to build a pricing model for my software with 2 extremely different cost structures...
Maybe the answer is both the offline desktop processing / upload to a server via scheduler for small to medium sized clients, and then offer the option of a dedicated server for enterprise level clients at a much different price structure... That way the contract covers the Server costs and the load is controlled because of single client use.
You made two comments that I think are especially important:
1.) "My priority is the speed and stability of the downstream client facing interactive software."
2.) "What about the cost side of scaling server software & high end hardware to handle an ever increasing resource load"
Server is scaled by adding more cores or additional worker nodes which allow you to process more workflows at a given time. The cost of this scaling depends on a few factors, but given what you are trying to accomplish I think that it would be best to have a conversation with your account executive who can loop in someone from architecture. I think that will help you understand what it will take from a cost perspective as you look to grow over time which I am thinking is probably the biggest priority for you.