What is our primary use case?
As an end user and a trained engineer working on field development, I am required to use a Linux-based system for all aspects of our work. This includes everything from logical design to design verification, and physical design, all the way to integrating data into the silicon database at the foundry. Since all of this occurs in a Linux environment, I must ensure we have the right platform in place. The performance we achieve with the tools we use can vary significantly across different platforms. Additionally, the support provided by these platforms is crucial. In the field of silicon design, we rely heavily on electronic design automation (EDA) tools, which are continuously being enhanced. As this area evolves, it’s essential for our operating systems to keep pace with the migration of the latest tool versions. If I become stuck with an outdated version of the OS, it can adversely affect my productivity and the quality of my designs. Therefore, I need to be reasonably familiar with various operating system providers and understand the pros and cons of each. This includes comparisons between Red Hat, SUSE, and Ubuntu, which is essential for meeting my requirements.
What is most valuable?
Since it is widely used, I believe the knowledge base is fairly good. In my own organization, which has three vertical companies, two others were already using Red Hat Enterprise Linux (RHEL) for production. They were asking me to go with Red Hat Enterprise Linux (RHEL) unless I had a compelling reason to go to SUSE or Ubuntu. This indicates that the IT team within my company preferred Red Hat Enterprise Linux (RHEL) for support and documentation purposes. The company has been around for more than a decade, so familiarity might be one reason, or resistance to change may have been another reason to stick with Red Hat Enterprise Linux (RHEL). In my role as the design manager, I have not heard anything negative about Red Hat Enterprise Linux (RHEL).
My decision to go with Red Hat Enterprise Linux (RHEL) was influenced by three main factors:
1. The IT team’s familiarity with Red Hat due to its previous deployment in other units.
2. Competitive pricing, which was 25 to 30 percent lower than other options.
3. The perception that Red Hat offered long-term service pack support for an additional fee; something that other providers like SUSE may not have offered.
Ultimately, the first two reasons were strong enough for me to lean towards Red Hat.
What needs improvement?
To some extent, I am speculating, but at the end of the day, the main thing we care about is how the resources are getting scheduled and utilized. Without an external load-sharing application, the number of cores in our servers and the memory should all be utilized effectively. If they can do very good dynamic resource allocation, maximizing the number of cores and the memory without external applications, that would be beneficial
Additionally, this is not just for Red Hat Enterprise Linux (RHEL), but for any OS - I would really love to make sure that their security features are robust and getting updated regularly. I believe at a given point of time, they may be very good, but hackers are also improving their techniques. I would definitely expect Red Hat Enterprise Linux (RHEL) or any OS provider to constantly monitor, understand if there are any new vulnerabilities in their OS, and provide patches or fixes so that we are always guarded from any security threat because what we are developing consists of very important IPs that have to be protected from malware attacks.
The most important thing is that it has to be stable. If it is not stable and we have to reboot it because of something, that would be problematic. The kind of tools it provides natively is important. For example, if I am doing development, I want to have a checkout process. If they have well-developed documentation and the ability to work with the code itself, along with good support for developing, then the performance of the OS would improve. If I see that one of my runs for any workload is taking five days, I immediately question why it is not completing within a day. If the load sharing is not happening correctly, there might be switches or features that the OS provides that can help use more memory or similar resources. Being developer-friendly would be beneficial. One thing managers hate is nasty surprises, so even if something is not working in the OS, it should provide some ability for IT to observe potential issues three or four weeks in advance.
For how long have I used the solution?
I have only been using Red Hat Enterprise Linux (RHEL) for a short duration of time, about six to eight months because the migration happened very recently.
Which solution did I use previously and why did I switch?
I am working for a startup company. We used to use open source SUSE because that was kind of easy to use and we did not have to spend many dollars. When we reached the point where we had to go to production, we needed to ensure we were using something more reliable because open source is open source. When I go to a newer version or a production version of the OS, some of the designs we are developing will be around because our startup is focusing on accelerators for the cloud. Some of these can be around for seven years, 10 years, and beyond. Hypothetically, even after 10 years, somebody who is using our silicon can find a bug, and we are obligated to fix it through software or other means. If we do not have the OS support at that point in time, because 10 years is a long time, it becomes problematic. When we go towards production, the kind of analysis that I do involves determining how many years this OS is supported and whether they will support it for an extended period, provided I pay them extension money. I am an end user, and I try to look at the facets of the OS based on my current business needs.
When we were using Ubuntu, I initially found it sufficient for my EDA tools under the evaluation licenses I had. However, as I progressed into silicon design and needed to purchase production licenses, I realized that the older version of Ubuntu wasn’t adequate. The question arose: if we were to upgrade to a paid version of the operating system, which one should we choose? I conducted some research comparing Ubuntu and Red Hat, and ultimately decided to go with Red Hat. Once I made that decision, I simply needed to explain my reasoning to my IT team, stating that I wanted to upgrade the twenty or so servers I was using to Red Hat 9.1, or whatever the current version was at that time. They took over from there.
How was the initial setup?
We experienced some initial challenges when we moved to Red Hat, mainly due to the tools' versions. At first, we struggled to navigate these issues, but once I contacted support, they were able to resolve them quickly.
The maintenance is handled by the IT team.
Which other solutions did I evaluate?
Most of the studies that I did were between Ubuntu and Red Hat Enterprise Linux (RHEL). I did not check extensively on SUSE Enterprise.
I was inclined to choose Red Hat for a couple of reasons. First, the IT team’s familiarity with Red Hat was crucial since it had already been deployed in other areas of the organization. This existing knowledge made the transition smoother.
Additionally, I did not inquire about pricing immediately because, ultimately, my business unit would be responsible for the costs. I recall that the price for Red Hat Enterprise Linux was less than one lakh rupees per license per year. The annual cost might be around 1.2 lakh or slightly more, but it was certainly under that threshold. Furthermore, I believe that if we were to negotiate for a larger number of licenses, we might have received a better rate. Regarding the initial pricing I received, I remember it being about twenty-five percent lower per license per year compared to other options.
For my use case with EDA tools, Synopsys EDA tools' local AE team said that support in India is better for Red Hat Enterprise Linux (RHEL). Additionally, Ubuntu and SUSE support for 10 years, whereas Red Hat Enterprise Linux (RHEL) supports for 10 years plus an extended two to four year period for a cost. Since our chips will be in the cloud market for at least a decade or more, this long-term support influenced my decision.
What other advice do I have?
I would rate Red Hat Enterprise Linux (RHEL) an eight out of ten.
Which deployment model are you using for this solution?
On-premises
Disclosure: My company does not have a business relationship with this vendor other than being a customer.