Things I learned in 2015
Big Data is here, and it’s not only a buzzword or trend anymore. The use cases become more and more clear, and tools are developed in order to make the Big Data professional’s life easier. The true game is architecture – how to use the right tool for the right task and make them speak to each other.
In the Big Data world, the term ELT replaces ETL. It means Extract, Load, and do the transformation at the query level in order not to lose parts of the data that we might need later.
Big Data also means working with Data Lakes, which means storing lots and lots of different types of data in their raw format, and combining those dispersed formats at the query level. It also means lots of projects end up as Data Swamps, meaning that there’s lots of data, but no one knows what’s in the data and when or where it came from. That’s why products like Azure Data Catalog are developed.
Amazon and Microsoft are in a constant race in the cloud, when the feature set is pretty similar. Amazon has Redshift? Azure has SQL Datawarehouse. Microsoft has Azure AD? Amazon has managed Active Directory. The decision regarding which cloud vendor to choose is determined by price, corporate culture, employee skillset, data center latency, and the maturity of the products. Roughly speaking, Microsoft are doing a good job in bringing their old friends (mostly corporates) to the cloud, while Amazon controls the startups and non-Microsoft oriented companies.
Storage in the cloud is a tricky thing. In Amazon, for instance, you get IOPS that are 3 times your storage size if you work with the free genral-purpose SSD tier. You can get past this number if you have bursts, but then Amazon will “punish” you, so that on average you’ll be back to the 3*storage size IOPS. In Azure, you have to stripe a few disks together to get more IOPS.
Machine Learning and Data Science are on fire, even though barely anyone knows what those terms are. If you’re among the lucky ones who know their way around it, you’re in a very good spot.
SQL Server 2016 is going to be a great version. It will bring In-Memory OLTP and Columnstore to the next level in terms of ease of use, ease of migration and performance. Execution plan lovers get cool features like Query Store, visual execution plan comparison and Live Query Statistics (visual presentation of row flow between the operators in the execution plan). It also brings a very elegant integration to Hadoop (with Polybase) and with R. However, Microsoft aren’t the only ones doing it, and I wonder how many people will actually connect their SQL Server instance to a Hadoop instance and start querying it. Only time will tell.
SQL Server 2016 still doesn’t supply a true scale-out solution. Microsoft APS and Azure SQL Datawarehouse are very impressive technologies, but they still don’t seem to match competitor products like HP Vertica.
Windows Server 2016 is moving to core based licensing, and to say the truth, this is pretty frustrating. Sure, it’s the “right” thing to do, but is it the smart thing to do? I believe that two main issues made lots of customers migrate from SQL Server to other platforms – the lack of a true scale-out solution, and core based licensing (that started in SQL Server 2012). Does Microsoft want the same issue with Windows Server?
MySQL is a nice database to work with. The principles are the same as in SQL Server, but the tools around it aren’t as mature, useful and comfortable as in SQL Server, and I’m not very impressed with how it works on RDS. However, partitioning is more straightforward in it and has more partitioning type than SQL Server. Also, Partition Elimination? It’s called “Partition Pruning” in MySQL.
Microsoft finally developed a good browser. I ungraded to Windows 10. It’s good, but the best thing is Edge, the new browser. It’s quick and pretty, and finally stands against the other browsers in the market. Well done Microsoft!
If you want something from your audience, just ask. We currently stand on more than 2000 listeners per episode of SQL Server Radio. This is awesome! However, we always want to help more people and show them how fun it is to learn through podcasts. One of the main traffic generators in podcasts is iTunes reviews, ratings and subscribers. So if you love the show, please consider leaving us a review and rate us, so we can help more people have fun. We’ll appreciate it a lot!
You can by doing the following steps:
Go to the SQL Server Radio page on iTunes
Click “View in iTunes”
Click “Ratings and Reviews”
Rate us (the more stars, the better)
Click “Write a Review” and type it.
What got you here won’t get you further. You’ve built a successful business/career and got to a good place. If you’re happy in that place, that’s fine. But if you want to keep growing (either as a company or in your organization/career), you’ll have to work harder/smarter, and most probably in a different way. This relates to companies, but also to us as individuals. Doing it is very hard, and most people don’t do that. But if you look at very successful people and companies, you’ll see that they keep changing and improving all the time.
Focus, and don’t multi-task. This is relevant in both the strategic level and the day-to-day level. On the strategic level, going after too many directions and business opportunities in parallel means you don’t have enough time to focus on each one of them, meaning you don’t finish working on any of them. And with that, instead of having one finished task that you can learn from and improve when working on the next ones, you end up with many unfinished tasks and business opportunities that didn’t produce anything yet. On the day-to-day level, it means not doing many things at once, but rather focusing on one specific task at a time. One tactic for doing it is blocking time in the calendar for the task and working only on this task on that time frame – no phone, no social media, and no other distractions – only the task. It’s still easier said than done though. Writing this blog post, for example, took me much more time than I wanted, partially because I switched between it and other stuff I had to do.
Want to stop trading time for money or increase your hourly rate as a consultant? Either start developing products and sell them online (courses, eBooks, software, etc), or get up the value chain. For instance, instead of being told to implement an AlwaysOn AG solution, make sure to be in the room when there is a discussion about HA, DR and Business Continuity.
Have a great year!