What's Good For the Goose...
My late maternal uncle (a retired Catholic diocesan priest) never believed in political polls: "I've never been asked to participate." Well, neither have I, but I have a decent background in statistics and probability. I know if the sampling methodology is solid, if the sample size is large enough, it's extremely unlikely your sample results will skew widely from the "real" mean.How then do we explain anomalies like Trump's unlikely 2016 electorate vote victory? Well, RCP rated Clinton with a 3 point lead in the popular vote, which she won by 2 in the election. And if you check the key 3 states (WI, PA, MI), the latter two were rated toss-up, and only Wisconsin, won by Trump by less than a point, was rated as lean-Clinton (by 6 points). Those 3 states had 46 votes of Trump's 306 and he needed 270 to win. Technically he could have lost PA and MI but needed to win at least one of the 3 states he won by an eyelash each. Keep in mind only Clinton in a 1992 3-way race won by a lower popular vote plurality and you have to go to the nineteenth century to find a POTUS winner who won with a lower percentage of the popular vote. I can't explain the Wisconsin discrepancy, but this post describes likely methodological issues.
Now I've created/administered a lot of questionnaires or surveys in my academic career. For all these favors I asked from study participants (of course, I thought they were interesting endeavors), I myself had never been targeted. Until now. The National Science Foundation sent me a couple of pieces of mail, one of which looked like some kind of donation request. No, it turned about to be a survey of doctorate recipients, and I had been randomly/scientifically selected. It was interesting but odd at points; for example, they asked for my starting date for a prior position, and then asked a question later which predated the start date. A lot of it deal with the usual demographic questions along with salary and functionality of work, relevance to my doctoral studies, any supervisory experience, etc.
I'm not sure how representative my information is. I probably have made 10-15% below market. For some odd reason, I've rarely had a line supervisory position; when I worked for an Equifax company years back, I had a much older data administrator working for me who basically ignored my assigned priority. He was later stripped from me by a manager who thought I wanted to fire him. I was stunned by outright insubordination. To be honest, I didn't need the hassle. I've served as the tech lead on several small/larger projects. I've often had informal authority over DBA's and developers; I've had senior managers, civil servants, and clients ask me for advice, assign me to special projects or as a mentor to other DBAs, or loan me as a resource. Oddly enough, I haven't been used much in talent acquisition, despite teaching hundreds of students during my academic career. It's not clear why; maybe they were concerned I might dominate the process, challenge their decision authority or scare off applicants. The last one I recall being on the acquisition side was on a project for a Chicago government business unit. I remember interviewing this Indian project DBA candidate, and it was clear from just general questions (no trick, overly detailed or obscure questions) that he was unqualified; I tactfully told him someone would get back in touch with him. He said, "Oh, I'll see you Monday. I have a 2-week minimum contract." What the hell? What was the purpose of having me interview him? I didn't have to work with him; he was someone else's headache, that guy bitched about him from day 1, and they let the guy go after 2 weeks. If they had waited for my feedback, they wouldn't have made that mistake.
Did my profession have anything to do with my doctoral program work? Not really. My dissertation chair, Richard Scamell, did teach a couple of database classes I took. UH had an abysmal DBA, so bad we all had to take incomplete's the first semester because the Oracle database was mostly down, not enabling us to finish our projects. That was a motivation years later for reinventing myself as a DBA after I was forced out of academia in a recession (even then it took a couple of years to get a break; I had to find a developer job in the interim. I got a low-paying starter job as a DBA at an EPA lab in Chicago). I did do a research project where I developed an SQL (the lingua franca of relational databases, like DB2, Oracle and SQL Server) minimal manual. My UH PhD student office mate Minnie Yen used the manual with one of her students at the University of Alaska Anchorage, and we later published our findings.
Microsoft Wins a Lucrative Cloud Computing Over Amazon
I can't go into detail here, except to note I recently worked a temp gig where a government agency is in process of transitioning apps on Microsoft Azure (cloud computing). I will simply point to a Microsoft infographic which claims over 80% of federal agencies have deployed applications on its Azure platform.Now for those readers with a limited IT background, cloud computing refers to maintaining servers running client applications and databases in external (vendor) data centers accessible through the Internet. Quite often these data centers have redundancy in the infrastructure minimizing risks to failure including geographic risk.
I gave an example of this in a prior post: I once did a DBA gig at a NASA site. Prior contractors had implemented a Data Guard (zero data loss replication) solution where the changes to a primary database/server are propagated to a standby clone database/server. So if something happened to the primary database (e.g., a hardware failure), the DBA could make the standby database primary. The problem? They deployed the two servers IN THE SAME SERVER ROOM (even the same rack). I repeatedly advised the government civilians/civil servants this was absurd. For example, if the server room got flooded, had a fire or whatever, the standby would likely be subject to the same risks. A server in another state/region would control for that risk. But government management just ignored my warnings.
Now Oracle, for which I once worked as a senior principal consultant (the highest non-managerial rank), can and has been a pain in the ass to work with, especially with uncooperative, arrogant, and/or incompetent analysts. (I could write pages of examples, beyond the scope of this post, but one recent example makes the point, probably one I've mentioned before. Oracle had recently published a comprehensive ODA patch. I reported a technical issue with the new patch; this analyst argued that Oracle hadn't released the patch, I must have installed "beta software", and they don't support beta software. He was in a state of denial. It wasn't a close call. Oracle's internal communications were fucked up; it was just he was in denial: I even detailed step by step how I downloaded the patch, and he just refused to admit he was wrong.)
Some additional context. cloud computing is a form of leasing computer capabilities. It's a variation of the computer time-sharing industry where I worked when I first moved to Houston. You don't need to worry about infrastructure or staffing its management. In Microsoft's case, they have two types of service PaaS (platform as a service) and IaaS (infrastructure as a service). One variation of PaaS is DBaaS. In the IaaS approach, the client retains more control over things like patching. You create a VM (virtual machine) and install the application or database software. So Microsoft provides the VM and the hardware/network support, but it's more of a DIY project and often you typically bring your software licenses to the cloud. In the DBaaS approach, Microsoft provides the software, will handle relevant patching, etc.
Apparently Microsoft Azure at one point provided Oracle as DBaaS. What's also relevant to this discussion is that Oracle has a competitive cloud solution, called Oracle Cloud. I heard from a civilian that Oracle competed on the cloud computing contract, lost their bid, and are protesting the award. But right now Oracle is competing with an aggressively-priced Azure SQL (DBaaS), a subset of Microsoft's competing SQL Server.
But there are two practical issues from an Oracle IaaS solution that I saw. First, it looked like 2 years ago, Oracle announced licensing changes that seemed de facto double the Oracle licensing/support charges to Oracle Cloud competitors Microsoft Azure and AWS (Amazon) vs. deploying on Oracle Cloud. Second, I'm on a conference call with Oracle and civilians. I asked one simple question that caught Oracle completely off-guard: why don't you have (Oracle's latest version) 19c in Azure Marketplace (like 12.2 and 18c)? The Oracle guys weakly tried to point the finger at Microsoft which publishes the images. What's worse, they NEVER followed up after the call. This was an issue because Oracle is scheduled to desupport the other two versions by June 2021. It seemed that Oracle's goal in the call was to promote the interface between Oracle Cloud and Microsoft Azure so we could implement our Oracle database on Oracle Cloud and the application VM on Microsoft Azure could connect to it. Well, knowing the clients, that wasn't going anywhere; at best, they were in a state of denial.
I was recently told by the civilians that management basically wasn't willing to spend a lot of money on Oracle licenses and support when they could get the functionality they need on economical Azure SQL. In effect, the burden of proof would be on me to justify the high cost. SIGH! Oracle is a technically superior, DBA-friendly product, but I had no information on scalability, etc., to make that argument. So it was clear it's time for me to move on to another Oracle shop. My agency had told me they had a contract for Oracle services. They will do whatever the government wants for the opportunity of a follow-up contract.
Of course, on an unrelated but topical note, the DoD awarded its JEDI contract to Microsoft over AWS (Amazon), potentially worth up to $10B. Of course, Trump has a petty feud with Amazon CEO Bezos, who of course happens to own Trump-critical Washington Post. You would hope the decision wasn't based on petty nonsense. Given Microsoft's already impressive government client list, it was probably a reasonable decision.
