Hacker News Reader

Comment by MichaelGG | Hacker News Reader

Take, for instance, SQL Server. It's become quite an interesting and capable platform over the last decade. It used to be that Microsoft took the stance that hardware improvements were the customer's benefit, so they'd license per socket, full stop.

They used to mock Oracle with their licensing depending on what kind of CPU and how powerful it was.

Latest SQL Server? Yep, you pay depending on the CPU vendor (AMD or Intel) and the type of chip.

View on HN

Just use something open-source - you'll get the Awesome [1] edition on any kind of chip.

I never understood why some people prefer to buy licensed stuff from Oracle or Microsoft. Do you like pain?

[1] http://www.codinghorror.com/blog/2009/07/oh-you-wanted-aweso...

I never understood why some people prefer to buy licensed stuff from Oracle or Microsoft. Do you like pain?

"nobody ever got fired for buying IBM equipment"

Right, there is more than one sort of pain. There is the pain your developers feel when they work with something painful, and there is the pain managers feel when they have to justify "not buying IBM" when something goes wrong (something will go wrong regardless, that is the nature of the beast).

So people "buy IBM" because they don't like pain.

The only good open source SQL RDBMS I'm aware of is Postgres.

The moment Postgres has the feature range of SQL Server (SSIS, SSRS, SSAS), supports the same HA features without hassle, and is generally even half as easy to manage, I'll switch. Meanwhile, life's too short for me to want to become a "real" DBA.

SQL Server has some extra tools, as you've pointed out. You can probably find open source tools equally as powerful (and, by definition, more flexible) to perform those tasks. Granted, you may have to invest some time finding and learning those tools - but really, that's just life. Sometimes it's worth putting in the effort [1].

I've found that the main difference is how you work as a developer. If you're happy using a clicky interface to get your work done, more power to you. I need to be able to script everything I do, because I don't like to repeat myself. The rigmarole of restoring a db from a backup in SQL Server is just insane compared to the 4 second command I could issue at the terminal for postgres.

I know it's going to sound like a total exaggeration but I literally can't use MS products for more than a few hours at a time because of the pain I get in my arm from using the mouse. And trust me, it's not just because I don't know my way around - I spent years developing on that platform (and will never go back).

[1] http://xkcd.com/1205/

It feels like using the mouse is the tradeoff for using Microsoft products. I use Visual Studio and SQL Sever at work, and they are such a productivity boost, but I hate how often I have to use the mouse. The problem is many of the benefits of using their products perks would be difficult to navigate around with a keyboard.

At times it's frustrating how often you have to use the mouse, but other times it's so great compared to any alternative ways (that I know of).

Learn the shortcuts, it isn't very hard.

SQL Server gives you both. They have a UI for everything, and everything you do has an option to script. So you might go through the whole wizards for replication, then when done, save off a T-SQL script. To my knowledge, there's nothing that requires the UI; it's there just to make things easier. Then on top of that, they expose an API you can use for further automation, if T-SQL isn't to your liking.

Your example is particularly odd: RESTORE DATABASE bla FROM DISK = 'file.bak' The docs are quite comprehensive, with lots of useful examples[1].

MS deserves some flak for UI-only, but SQL Server isn't one of those products (at least since 2005). MS products in general are also becoming much more scriptable, as Powershell access is becoming a requirement for new products.

I prefer a clicky interface for things I'm not going to do often (like configure a new cluster) or things I don't want to commit to memory or have to look up. Anytime I need to repeat, I either read docs and construct a command, or if it's a complicated thing, I'll use the wizard to generate a base script and modify as needed.

1: http://msdn.microsoft.com/en-us/library/ms186858.aspx

It's not all roses. Have you ever tried scripting and versioning SSIS for example?

Probably not.

Which is why we use Spring Integration...

Yes one thing that always amused me at microsoft was how often they would talk about how much better their tooling experience is. It's not. Being able to go into roo or ror or whatever else and generate huge amounts of scaffolding and get tons of upfront work down with out only a few keystrokes beats the socks off of an obtuse and clunky sql and ide interface where some portions work well and some portions look like remnants of the early nineties.

After googling all the acronyms, I would like to know what features you believe SSIS, SSRS, SSAS and HA give you can not get from Postgres with drop in replacements.

Except maybe for that everything is already integrated and "just works".

What I'm asking is if you can specify some elements of the stack you mentioned, that is hard to replicated or unique to SQL Server?

Well first, integrated and "just works" are pretty high on the importance scale. Searching for PG clustering gets me a wiki with a matrix of different solutions, with apparently the main suggestion being DRBD. So just to start off, I've got to evaluate a number of third party products or start quizzing community members to figure out which is the most likely to survive or put me in a good position.

They might not be "hard to replicate" or unique, but having great options in-box, that are extremely simple to configure - that's worth a ton. If I was running a large-scale operation and SQL Server was costing me millions a year, things might be different.

Everything is awesome until you stop doing general stuff. Once I started digging into more complex MDX queries in SSAS (calculated fields) the story started to be quite different. I have been doing Win dev since 1993 so I could tell you a lot of stories...

Because there's not an OSS equivalent to Oracle?

That article from codinghorror is absolutely brilliantly written. But it's true, I wish there was a counter somewhere about how many millions/billions were lost by web developers trying to make things look like a webpage inside IE. I will call the new version IE too because I know that it will suck, it will. They have good programmers, they bring pc to the masses, I give them that but they have no clue about design or user experience and intuitiveness, they never have and never will. It's just a coorporation with strict bureaucracy where a good developer gets absolutely nothing done.

I thought it was interesting when Microsoft bought Skype. Skype built their whole backend on postgresql and were one of the major contributors to the clustering and hot failover features as well as putting up some nice management utilities. that's why o think the Skype purchase was so strategic. they not only got cross platform messaging with video and phone, they got the primary developers who were adding top end features to an open source project.

Reply on news.ycombinator.com