Planning and architecting Tableau Server

Kyle Roemer

Kyle Roemer

Have you figured out where Tableau fits into your BI shed as a tool for the enterprise? If you haven’t already, make sure you read my colleague Ronak’s post.

Now that you’ve decided to bring Tableau Server to your org, it’s important to properly plan and architect your new tool to best support your business. Here are some key considerations to think through before you get started.

How big are you going?

Whether you’re starting with a staggered deployment to key departments in your organization, or deploying Tableau Server to your entire enterprise, you’ll want to consider these inputs to estimate the demands on your server architecture.

How many concurrent users do you expect on your server environment?

Consider on average, but plan for peak; these numbers can be taken from your existing reporting environment. For example, 70-100 concurrent users can be supported by 8 cores (irrespective of how many machines you split the cores on).

Will your users be connecting live to your EDW, data marts, or Big Data appliances, or will they be extracting data?

It will probably be a mix, but plan for your typical use cases. For example, in heavy extract environments, you’ll want to split out the data engine but also make sure to generously account for RAM & Disk Speed (8 GB RAM/Core is a good rule of thumb).

Do you need to ensure the server environment is always on?

Consider that high availability requires a minimum of three machines (Primary/Gateway and Workers) and you should account for having a backup primary for failover support. True high availability with Tableau Server means you can support the same load on the server if any machine goes down.

Will your users be geographically dispersed?

Consider geo-specific server environments, and replicate “primary environment” content files between the other environments.

Plan for growth

If you’re deploying to a particular department within your organization, you should plan for other departments jumping on board. In short, plan for growth!

As you add users onto your server, plan to add additional servers. You’ll also need to add VizQL processes—a big component to concurrency.

If you’re deploying Tableau Server out to an organization of over a couple hundred users, you want to make sure you go with core-based licensing. If your deployment is smaller, named users will work. If you’re looking for a hosted solution, Tableau Online is a great option!

Will you go virtual or physical?

Does your department or IT team have a BI standard for VMs or physical servers? When it comes to utilizing VMs or physical boxes for Tableau Server, you should know both will work well at any scale.

It’s worth considering whether you will scale your server out or up. Scaling up is done best with larger, physical servers with the CPU & RAM to handle the added load, while scaling out is preferred with smaller VMs. There are advantages and disadvantages with both approaches, but when deploying out to the enterprise, I recommend scaling out to reduce Tableau Server process competition and ensure process redundancy. (8 cores gives an optimal core/performance ratio, but if you’re going with user-based licensing a 12 core box [many servers come with 12] is a good choice.)

Just because you build it doesn’t mean they’ll come

Don’t fall into the classic IT trap of believing in the Field of Dreams mantra. Yes, this may work in the app and social space, but why not set yourself up for success here? One of the most important components of a successful Tableau server deployment is ensuring your user base is versed in how and why they should use this tool. Train and enable your users to get the most functionality from Tableau Server, and you’ll see increased adoption and a banner of you hanging in your org’s Analytics Hall of Fame.

If you’re planning and architecting Tableau Server and need expertise to make sure it’s a smashing success, please reach out to our team of Tableau rockstars.


One Response to Planning and architecting Tableau Server

  1. Pingback: How to improve the performance of your Tableau dashboards | The Slalom Blog

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: