That's a common question in IT and my opinion on certifications has changed over the years. I recently got my [[Vertica certification]] and in the process I've been re-evaluating the value I find in certifications.
My First Certification
...was my Microsoft Certified Professional (MCP) in 1997. I worked at a large "shop from home" television channel as a help desk guy. The company allowed me to take a Windows 95 class and they paid for the certification. I already had my B.A. and I didn't find a lot of value in having an MCP.
Microsoft Certified Trainer
I didn't stay at that company for long. I went to college to be a high school teacher but realized I hated teaching kids. So I thought I could teach adults SQL Server, databases, and programming...something I did to help put myself through college. I was hired by a MS Gold Partner and soon became an MCT. Within a matter of a few months I held my MCSE and was working toward my MCSD. Being an MCT for various products requires that you pass the exam for those products. This usually meant learning just enough to pass the test so I could teach classes and earn my pay. My breadth of knowledge was diverse, but had very little depth.
As the aphorism goes, "those who can't do, teach". I was teaching Exchange Server after just barely passing the exam the week previous. As long as I stuck to the courseware I was fine and students loved me. I even won "Instructor of the Month" for all of the United States in April 1999. But if I veered from the courseware, or a student asked a moderately difficult question, I looked foolish. This really wasn't what I wanted out of my career. I felt like a shallow fraud.
It was around 1999 when I first noticed the ads on the radio touting, "Get your MCSE and make $70,000 per year your first year. We can help you!" These were basically bootcamps where you were given the answers to the exam questions and were allowed to briefly work with the technology. They had similar bootcamps for Java Jocks. This was the Dot Com Era after all. The market demand for technologists was never higher. This disgusted me. My training company was also a Prometric Testing Center and I saw a lot of test takers come in to take the tests to get their certifications. It was common to see people in the parking lot studying their "brain dumps" that they downloaded from the nascent Internet. TroyTec was becoming hugely popular. I saw tons of old TroyTecs laying around near the exam area that testers left behind as they were trying to cram just before taking a test.
These people were "Paper MCSEs" and it put a sour taste in my mouth for a decade regarding certifications.
One Particularly Disgusting Experience
Every Prometric exam room had a security camera on the wall plainly in sight of every test taker. Our office administrator had a CCTV sitting prominently on her desk that she used to monitor the examinees. Obviously this was to discourage cheating. In fact, if you took an exam you had to sign a little waiver indicating that you would not cheat or sell the questions to a brain dump site.
I came back from lunch one day and our office admin asked me to look at the TV with her. Right there, clearly, was a test taker thumbing through a folder of notes looking for answers. And there was the camera right over his head. We often saw people cheating...you could see them looking at note cards they had up their sleeves for instance. The office admin asked me what she should do in a case this brazen.
My response still sickens me.
I told her to ignore it.
If you accuse someone of cheating it can wreck their career. What if, somehow, he wasn't really cheating? What if we accuse him of cheating and somehow this blows up in OUR face? I would prefer not to be involved AT ALL in these scenarios. As I saw it at the time, if you cheated and got away with it...good for you. You're a fraud and you won't last long at your job.
These cheaters were the rule, not the exception, during this era. I viewed a lot of it. It disgusted me and jaded my opinion of anyone who had a certification but no real-world experience.
You could successfully cheat for just about any MS exam, but Networking Essentials was the hardest. It was the widowmaker. I saw guys take that test 4 or 5 times and still not pass. "Net Ess", as we called it, required you to actually DO networking. You couldn't just memorize brain dump answers. You had to conceptually understand routing. And you were asked to do math involving subnetting and supernetting. You even needed to know about "plenum wiring", which I had run for 4 years as an intern in college. The test was still multiple choice so it was possible to cheat, but it would likely take you a few tries.
If you wanted to know if an MCSE was of the "Paper" variety, simply ask them how many times it took them to pass Net Ess. Or ask them a question from the test.
In 1999 MS released SQL Server 7. This was a radical departure from SQL 6.5 on many levels I won't rehash here. I had limited experience with 6 and 6.5 and IMHO the industry scuttlebut was that SQL Server was only a "departmental" database...not able to scale and too prone to crashing on Windows NT. With SQL 7 the industry's views seemed to change overnight. SQL Server was now considered a suitable database for these new e-commerce apps that were becoming popular, and mission-critical.
Shortly after SQL Server 7 was released MS announced the availability of the MCDBA certification. Since I worked for a Gold Partner I was privy to "inside information". I learned that the two SQL test requirements (Administration and Design) for the MCDBA wouldn't be quite so easy anymore. Was Microsoft finally going to have stringent testing. Would there be no Paper MCDBAs? Although still multiple choice the questions would be more scenario-based. Gone would be pedestrian questions like "Where is the data for a view stored?"
I begged my employer to let me take the tests before they were GA. I studied and passed both exams on the first try sometime around August 1999. I was probably one of the first MCDBAs outside of Redmond. The tests were not easy and I remember seeing many candidates fail the exams for the next few months. Around this time MCSEs were everywhere but did not command much of a premium in the market. But an MCDBA did.
The Windows 2000 "Upgrade" MCSE
I soon tired of being a technical trainer and became a full-time data professional in late 1999. It was easy to forget my former career.
When Windows 2000 was announced MS also announced that the Windows NT MCSE would be retired. If you held that "version" of the MCSE (which I did) you could no longer call yourself an MCSE without taking an upgrade exam. This pissed me off. This was like having my B.A. removed...I earned it I should get to keep it. I wasn't using any of the new features of Win2K, why should I upgrade? I still felt that there was a bit of value in having the MCSE logo on my resume so I decided I would at least attempt the upgrade exam. The upgrade exam was available until January 2002 and I scheduled my exam for the last week of December. I didn't study and I barely passed.
What really pissed me off is MS reversed their decision a year later and let everyone keep their MCSE designation even without the upgrade exam. It is this kind of attitude by MS which is why I'll likely never hold another MS certification again in my life. They compounded bad decisions upon bad decisions. I talk to a lot of veteran MCSEs who still are not really sure if they are allowed to use the MCSE logo or not based on these ridiculous rule changes.
In 2002 I was using Oracle every day designing ETL packages. My employer sent me to training so I really just needed to pass a few tests to become an OCP. I passed each exam easily. But OCP exams are much like MS exams...very pedestrian questions with obvious answers. I'm competent with Oracle, but frankly my skill level isn't great. I guess that makes me a Paper OCP. The fact that I aced every exam kinda indicates to me that an OCP certification, like an MCSE certification, is not a good measure of a potential new hire.
The worst aspect of the OCP tests is that you don't need to know ANY PL/SQL. Every day I encounter Oracle DBAs that cannot write even rudimentary PL/SQL packages. I strongly believe this is because the OCP exams do not stress this.
It was another 4 years until I got my next certification. I was working as a data architect at an ISV and our hosted environments were having HUGE I/O problems. I knew at least the basics of SANs and storage and I/O but I was having problems communicating my observations with the SAN engineers. They were denying their SANs had any problems...and I knew MOST of our problems were indeed the SAN. I began to read anything I could about EMC, IBM Sharks, or anything storage-related. After a few months I was able to articulate, with irrefutable proof, that we had SAN problems. I even got the SAN engineers to admit it. It took a few more months before our problems were fixed.
Maybe I could pass the SNIA SCSP. I took the test and aced it the first time. No one paid me this time, I simply wanted the certification to prove to myself that I knew the technologies as well as I thought I did. Even though I aced it, I guessed at every question involving tape libraries and NAS...I didn't feel those subjects were germane to a data architect. But apparently they are very important to the fine folks at SNIA.
I think the SNIA SCSP is the most valuable certification I have. It is an industry-group certification, not a vendor certification. This means that it is vendor-agnostic. You are tested on concepts and theory vs tools. With MS and Oracle exams there is clearly a push to have candidates learn the newest features of the newest release, even if those features have a limited applicability to most practitioners.
Vertica, MySQL, etc
I recently took my Vertica exam and passed which makes me an HP ATP. I've been using non-traditional DBMSs a lot lately, especially Vertica. I had no formal training in Vertica, I merely picked it up by reading and building my own test servers. I even have a Vertica series of blog posts where I'm disseminating some of the key Vertica concepts that initially tripped me up. I took the test to prove to myself that I knew Vertica and to determine what areas I needed improvement. In other words, I took the exam for ME. I'll put it on my resume not because I value the certification, but because I do have valuable Vertica experience.
Unfortunately I think the Vertica exam is a complete waste of time. Here's why:
This is just another vendor exam that focuses too much on the tool and not enough on the concepts.
I find little value in most certification exams. Vendor exams are almost worthless. There is too much emphasis on the newest features which are likely not being used, yet, in most shops. And there is too little emphasis on foundational knowledge. Industry certifications (SNIA, PMI/PMP) are much more valuable. A thorough understanding of the concepts is required, and those concepts are transferable to most vendor implementations.
I especially like the approach that EMC is taking. They offer a Data Science Associate certification. This is a holistic "architect" certification. You must know a little bit of everything in data lifecycle management. Design is covered, relational concepts are covered, querying in multiple languages is covered, as well as storage (this is an EMC certification after all). Even though this is offered by a vendor there is nothing vendor-specific about it.
You have just read "[[On Certifications]]" on davewentzel.com. If you found this useful please feel free to subscribe to the RSS feed.
Dave Wentzel CONTENT