Watch this Ask the Expert session to hear examples from customers who have been through a migration from SAS 9 to SAS Viya.
Watch the webinar
You will learn:
- The benefits for IT and SAS users of a more modern cloud platform.
- The resource and platform requirements to support the modernization.
- The key benefits and differences in a new, modern approach for analytics.
The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.
Q&A
The resource list below refers to Viya 3.5 - are there similar comparisons for Viya 4?
At the top of the page, there is a drop down where can select the version of Viya you want to compare to SAS 9.
Can you please provide more information about hosting SAS on-prem?
Here are few things to get you started: https://communities.sas.com/t5/SAS-Communities-Library/Deploying-SAS-Viya-using-Red-Hat-OpenShift-Gi... https://go.documentation.sas.com/doc/en/itopscdc/v_025/itopscon/n1ob03chiyninhn19sm1ssfgss7w.htm
Are there options to buy a hosted Viya environment from SAS (or 3rd parties) so that they do the platform admin work / cloud work?
Yes, SAS provides hosted managed services. Please reach out to your Account Executive and they can discuss this with you.
Is Viya 4 available for on-prem, since it’s available for Cloud?
Yes, we have an OpenShift option. https://communities.sas.com/t5/SAS-Communities-Library/Deploying-SAS-Viya-using-Red-Hat-OpenShift-Gi... We are doing an Ask the Expert on this topic on July 19th.
Which cloud provider is best for Viya 4 when it comes to ease of deployment and infrastructure maintenance and how does that choice affect third party database connection like Sybase Oracle, PGSQL, MSSQL, etc.?
I cannot comment on which provider is best. It really depends on the customer and their requirements for a cloud provider and IaaS solution. We have been very successful in migrations on both AWS and Azure. SAS is becoming more tightly integrated with Azure as well, so if any of those capabilities meet your business needs it’s something to consider. The second part of that question. How does it affect third party database connections like Sybase, Oracle, PGSQL, MSSQL, etc? That also depends. If you're going to migrate your data to the cloud, you need to make sure that they are supported there. My suggestion is, as much as possible, to use the native databases within the cloud like RDS, or other managed database services that the cloud providers have. You can also just set up instances in the cloud to support those databases. Regardless, SAS still has the same capabilities of accessing relational databases and other file types, so that's not a limitation that you're going to now encounter because you move to the cloud.
How can someone migrate from an older version of SAS, such as SAS9, to SAS Viya? Is there a difference between older versions of SAS to SAS Viya migration, such as from SAS9 to SAS Viya migration?
It depends. I would start by assessing and inventorying what you want to migrate. Most migrations we have performed are on 9.4 or Viya, so the standard tools apply for support. Previous versions may require manual processes for content assessment and code migration.
How does one upskill from SAS 9.4 Administration to SAS Viya Administration?
Yeah, that's a great question. My starting point for the answer - there would be 2, if you already have a really solid Linux administration type skills you can upskill them through training and hands-on experience with managing a Kubernetes cluster. The transition of Linux commands and skills overlaps with that of Kubernetes. It's similar types of back-end processes and interfacing with the operating system itself. There is training out there on Docker on Kubernetes and on containers in general that I would recommend, and then also looking at how that would interface with whatever provider, whether it's an on-prem provider where you're deploying your own Docker and Kubernetes clusters. I think that being able to learn Kubernetes is the place that I would start in terms of upskilling and training your administrators. Assuming they already have Linux skills.
What kind of SAS code change is expected when it comes to switching from GSUB to CAS?
It sounds like this question is coming from a grid user. One of the things about SAS is if you've deployed it in a distributed fashion, which you know is one of the major benefits when you're talking about large volumes of data and data processing, you don't have to put commands in there to distribute your workload. If you're talking about GSUB, where you might have certain nodes in your grid cluster that you want to use for different purposes, you use the GSUB command and you submit code to a particular node, or just to the cluster for it to determine where to go. One of the nice things about CAS is you don't have to be prescriptive that way, so you would just submit the code and then that code gets distributed across the cluster during processing. It's more of a parallel processing type environment than a, let's call it, work for workload management type environment. Hopefully that answers the question.
I would like to map for migrating SAS DIS jobs to Viya - what is the right mapping?
I think that's a similar question to what we were just discussing. DIS, Data Integration Studio, will provide some more information on what capabilities, in terms of migration, exist, and then what considerations you need to take and changes that you might have to make.
How different is SAS coding?
It's a great question. It doesn't have to be different at all, but there are capabilities that are available to optimize the code to run in bias. When I was previously talking about the road map and the migration strategy, you can program just like you do for the most part today in SAS. In SAS Viya, there is a comparable compute server within SAS Viya that will process. I think where you really want to change your code is when you have some of the recommended CAS capabilities for things like your longer running steps that you might have in your traditional programs or your CPU intensive steps or just trying to deal with larger data volumes. Hopefully that answers the question but doesn't have to be different.
Can Stored Processes be migrated - or when will this migration functionality be available?
Good question. In terms of migrating stored processes, there's a job execution flow that’s part of SAS Viya that we typically use to convert stored processes. It is very similar where you have URL you can call to pass in parameters. There may be some changes that you would need to evaluate in terms of how the parameters get to the store process or job execution flow.
Here is a good reference for how to migrate stored processes to the job execution server.
https://blogs.sas.com/content/sgf/2020/02/28/jobs-stored-processes-in-viya/
Is there Data Integration on Viya comparable to DI Studio and can we migrate DI Studio Jobs?
Yes. You migrate from DI Studio to SAS Studio (in Viya).
https://communities.sas.com/t5/SAS-Communities-Library/SAS-Viya-2021-2-Migrating-SAS-Data-Integratio...
https://go.documentation.sas.com/doc/en/webeditorcdc/v_007/webeditorug/n1c75how7oihsnn1lwbrp34vtw8k....
Do you typically see the data move to the cloud first, and then the SAS migration happen? What patterns do you see?
Yeah, a combination. I'm not trying to be vague, but we do see a combination. We have some organizations that we work with where they have a full cloud migration strategy and part of that is data residency and moving the data to cloud native databases. And if that's part of the customer strategy, then yes it makes sense to migrate your data first because that's going to impact not just SAS. That's likely going to impact other systems, operational systems, etc. However, we haven't run into many significant issues when the data is not migrated. I mean, there's some lag and latency, but there's cloud tools out there that help accelerate the access to your data that's on-prem. One way is creating VPN gateways. But then there's also accelerators from the different cloud providers to make a bigger pipe to get to your data. It really depends on your overall strategy for migrating. Are there certain databases or data types that you want to migrate sooner rather than later? To get the best processing, yes, you want your analytics as close to your data as possible. If you can migrate your data along with your journey to migrate SAS, I would suggest doing that.
Does this integration need to open additional ports and add additional certificates?
I'm not totally clear on what the integration they're talking about there, but if you're within the cloud environment, you know that load balancer typically the way we set that up. If you were looking at that architecture that we looked at a few minutes ago, typically the way we're going to set that up is everything is going to be private and not open, except for a pathway through the load balancer. That load balancer uses a certificate, it doesn't need a new one. It could be managed by the cloud provider, or it could be a certificate that your organization provides. But all any access goes through that load balancer. All only access from that load balancer can get to the environment. The servers themselves are set up in a private subnet and then you have a public subnet, and when I say public it could only be accessible through VPN or through your organization, whitelisted to certain IPS. That public subnet is how access is determined and everything else is behind your traditional Firewall.
What would you say is the value proposition for SAS Viya compared to more established Cloud Analytics platforms, such as Cognos/Tableau/Qlik, etc.?
When I hear a lot of those other tools, I actually think more visualization and more reporting. Some of them have basic analytical capabilities. I do think that the biggest value proposition of SAS and Viya in the cloud is that native capability of running in the cloud and integrating tightly with other languages and systems. And what do I mean by that? I mean the ability to provide more of that on the fly elasticity. The ability to provide a single integrated platform to incorporate. Leveraging a single platform for managing and executing their analytics and an environment that can truly scale in a cloud native way versus a more traditional way.
Does the content assessment tool work on Unix servers?
Currently supported platforms are AIX (UNIX), Linux, and Windows.
https://support.sas.com/downloads/package.htm?pid=2465
Given the necessary cloud skills for migration, do you see companies up-skilling their platform admins, hiring new admins with the skills, or buying / contracting a Viya environment from a 3rd party provider? In my experience as a SAS user, it's been a challenge to find companies where the IT is truly interested in investing in SAS platform management. I'm wondering if it would be easier to look to a 3rd party to host a Viya environment for us.
As a third party we have engaged with customers in this migration process many times and I think having someone that is tied into the enablement plans from SAS and works closely with various teams can benefit the customer. There are different remote managed services that are available to customers from 3rd parties, for both deployment and ongoing support of various SAS platforms.
Any training on containers or Kubernetes from SAS?
There are a few courses offered for SAS Viya Administration and Operations that provide high level training on containers and Kubernetes as it relates to a SAS deployment specifically.
How to migrate stuff from DIS to Viya? How does the lineage work here?
https://communities.sas.com/t5/SAS-Communities-Library/SAS-Viya-2021-2-Migrating-SAS-Data-Integratio...
https://go.documentation.sas.com/doc/en/webeditorcdc/v_007/webeditorug/n1c75how7oihsnn1lwbrp34vtw8k....
How are the custom transformations supported? Any other specialty on top of what we have in current SAS 9 env?
Per this SAS note, the transformations become SAS Studio Steps, which supports custom written SAS code.
https://documentation.sas.com/doc/en/sasadmincdc/v_024/calcontentmig9x/p0lui94y7l3fz5n1uzg6rgzgdyjy....
Are most people doing their own deployments or needing help?
Most organizations are engaging outside help for the deployments, but it depends on the SAS and non-SAS skills within an organization and the complexity of their current content.
Our on-prem environment is heavily invested on SAS 9.4 and we also have Viya 3.5. In the next few months, we will be moving to Viya 4. As we have both environments, how will the road map look like for our environment?
This is tough to answer without more information, but it is dependent on the type of content and in 9.4 and 3.5 and your goals for moving to Viya 4.
It seems SAS Viya is a next generation platform for past/current SAS. As far as you know, are there any pharmaceutical companies heading to SAS Viya?
Yes, there are pharmaceutical companies heading to SAS Viya. You would need to include an evaluation of validation, or other CFR type requirements in your roadmap.
Can you migrate Data Management Studio (DataFlux) jobs to Viya? These would be jobs written in the DataFlux Expression Language.
Not yet, but we are rapidly moving in that direction. We will soon have proc EEL that will allow that code to run in Viya and is necessary for us to build out the flows.
Does all data have to be in cloud or is a hybrid in cloud/on premise possible?
Many organizations have a hybrid cloud/on premise data solution.
Why is the SAS platform complex in this kind of infrastructure?
It is different. The complexity comes from SAS modernizing their software to provide a more scalable manageable infrastructure and platform as well as transform the way it interfaces with cloud native capabilities.
Recommended Resources
SAS 9.4 ad SAS Viya Functional Comparison
Relationship between SAS 9.4 and SAS Viya
Comparing SAS 9 and SAS Viya
SAS Viya Product Page
SAS Viya Highlights Release Show
Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars.
Click image to register for webinar