October 20, 2014 at 7:40 am
Of course our end users don't care where or how we store the data as long as they continue to get the information that they need.
We could store data in Magical Coconuts as far as the end users care. As long as they get their reports, the system is secure and backed up, they don't care.
At some point price may become an issue, so the folks on Oracle and DB2 may choose to switch to SQL Server and others may choose to switch to MySQL or Postgres...
October 20, 2014 at 8:09 am
I don't have anything against Oracle or any other RDMS, but I tend to always recommend SQL Server if for no other reason than it is the cheapest to support from a personnel standpoint. If your DBA leaves, there is a plethora of talent available and it doesn't take a rocket scientist to just keep things afloat until you can find the right long-term resource. Most small companies don't even require a true DBA.
* To stay on topic, I will say that my current company is switching to Teradata because the resources they do have seem to be incapable of properly scaling our DW using SQL. The "new" parent company seems to be quite comfortable with throwing mulah at the problem.
Aigle de Guerre!
October 20, 2014 at 8:28 am
MalD (10/19/2014)
Another consideration is, the correct tool for the job.. Having recently been involved with a mapping application was astounded in the the maturity of the Postgres implementation of Geography/Geometry functions. Yet SQL Server as a production system still, in my opinion, provides a data and development environment to answer most business needs. Without getting into a 'whos flavour is better' debate, they all have their strengths (and weaknesses). Beta or VHS .... like many points in database development... it depends.CodeOn
:w00t:
So very true. There are certainly some systems, like Neo4J, that solve problems in ways that might be very difficult with other databases.
However, you'd certainly have to weight the advantages of better performance with the overhead of training/learning/making mistakes on other platforms. Would I move to PostgreSQL over SQL Server if I ran a mapping application?
It depends.
If my company depended on it, I might lean that way. If it's one of twenty apps, then not sure. I'd have to really think about that. Either way, I'd have to experiment in a small way, maybe some POC that gets enhanced across months to determine whether it's a good move.
October 20, 2014 at 8:31 am
Yet Another DBA (10/20/2014)
courtney.smith (10/20/2014)
I'd argue against that. I don't think that expensive and difficult are really on the same wavelength / scale for comparison.What's so expensive about MS development? SQL Server Express still allows 10GB and is free. There are express versions of Visual Studio if cost cutting is the end-aim.
For a web-presence just get the licensing with the hosting.
In terms of development languages C# hands down beats java / objective c (developer trust-wise it is a different story). Even basic language features like LINQ, or even auto-properties.
I do agree with you. Its just that they dont want to listen, no matter how it is phrased, tried on so many other aspects. To them because they pay the wages they can define what shape the world is not matter what the evidence is. Sad to say, many business decisions are not based on logic but instead are based on whim, hearsay and prejudice.
I get that. Writing a check for licensing seems over the top, especially when you have to write a check for salaries either way. It takes some forethought and understanding to appreciate the lost opportunity costs when your development slows as people try to learn new skills on the LAMP platform.
That being said, so much of the platforms borrow from each other, that initial development might not be that much slower, but as you approach performance limits on your hardware and more complex features, things get hard. 'course, you can always buy more hardware to mask issues, and that might not be a bad plan in some cases.
October 20, 2014 at 8:49 am
Thanks, Steve, for this editorial.
To me it seems that cost is a big factor along with the relevant features of each platform. I think as far as relational database platforms, people can make cases for various DBMSs (SQL Server, Oracle, MySQL, Postgres), but as I understand it, Oracle can be really expensive to implement, followed by SQL Server Enterprise Edition installations.
What concerns me lately is the huge push towards NoSQL. I think NoSQL DBs have their place but I think too many people see them as a panacea that will allow them to stop using RDBMSs. From what I have learned of relational database theory, I think that's a mistaken path. I think relational and NoSQL each have places where they would be useful, so thinking NoSQL is somehow the solution seems misguided to me and will likely result in the reappearance of data anomalies. It can be hard enough to keep relational databases from being polluted with bad data and tables.
Also, I have read a couple of worrying articles that automation is increasing so fast that in 5 years many administration tasks may be replaced by automated procedures. I guess good DBAs already use automation, but it sounds as if the process is being escalated so that even setting up those procedures is going to be automated. Sounds like a shrinking landscape for DBAs (except for the analytical and troubleshooting things that humans still generally do better), but maybe I'm wrong. The only reason I bring up that part is if specific vendors move faster in that direction, it may influence platform selection of Oracle vs SQL Server etc.
- webrunner
-------------------
A SQL query walks into a bar and sees two tables. He walks up to them and asks, "Can I join you?"
Ref.: http://tkyte.blogspot.com/2009/02/sql-joke.html
October 20, 2014 at 9:09 am
Also, I have read a couple of worrying articles that automation is increasing so fast that in 5 years many administration tasks may be replaced by automated procedures. I guess good DBAs already use automation, but it sounds as if the process is being escalated so that even setting up those procedures is going to be automated.
The low-hanging fruit will be automated. And it should be. I watched a DevOps guy set up a new server system in OpenStack in less than 5 minutes. Given a script and an ISO, I can rollout a Linux desktop ready to go in 30 minutes, completely loaded with applications and totally patched. (Most of the time is waiting, so I can do other work..) My coworker can do the same with Windows. (Or we can just create a virtual machine for the task...)
However, complex issues will still require a trained, able mind to resolve them. Good code and queries doesn't write itself no matter how many fancy ORMs you throw into the mix. Good people still need to clean data and build ETL and processes.
October 20, 2014 at 9:36 am
When thinking of costing, we cannot forget what are ongoing costs, year on year, and what are repeated costs, upgrade after upgrade. If we remember this then sometimes the costs appear differently.
Also, there is a hard to discern value to providing the technological landscape that is desirable for the IT department. If everyone dislikes the selected software then I would imagine staff retention would diminish and vice versa.
Gaz
-- Stop your grinnin' and drop your linen...they're everywhere!!!
October 20, 2014 at 9:57 am
webrunner (10/20/2014)
Thanks, Steve, for this editorial.To me it seems that cost is a big factor along with the relevant features of each platform. ...
Oracle can be really expensive to implement, followed by SQL Server Enterprise Edition installations.
Very true. The gap is closing, however, and certainly lots of people have looked to move apps away from larger platforms (mainframe/mini) to x86/x64s to reduce costs. The same thing happens with people looking to leave Oracle/DB2/MSSQL for MySQL and PostgreSQL as well as Mongo, couch, etc.
What concerns me lately is the huge push towards NoSQL. I think NoSQL DBs have their place but I think too many people see them as a panacea that will allow them to stop using RDBMSs.
...
I think this is developers trying to ease their pain in the Object<->Relational mapping. They move to something that stores a document, an object, or JSON serialization, and makes things simpler. However, many of the issues crop up when you go beyond simple storage and end up turning some of the NoSQL solutions into documentdbs with SQL bolt ons.
We're very immature in this area (in general) in terms of sorting what problem domains fit better into RDBMS v other databases and how to then also translate the data into formats for reporting, querying, etc. I suspect that we will fight this issue for a long time.
Also, I have read a couple of worrying articles that automation is increasing so fast that in 5 years many administration tasks may be replaced by automated procedures.
...
I've been hearing this since the 90s. I don't believe it. We automate faster, but we add so many more machines and processes. We add complexity that can't be automated. What we do is raise the bar for knowledge. Log checkers, backup admins, these jobs are going away. You do need to do more in many cases. Especially as many smaller companies turn to services (SaaS) to meet their needs for things like email, docs, storage, etc. There still needs to be some admin, but certainly fewer of them.
However I think there always is a lot of work. Especially work that so many people complain they never have time to do (like tuning). Now is the time to make sure you can perform those other tasks.
October 20, 2014 at 10:00 am
Gary Varga (10/20/2014)
When thinking of costing, we cannot forget what are ongoing costs, year on year, and what are repeated costs, upgrade after upgrade. If we remember this then sometimes the costs appear differently.
I think we're seeing less of this. Companies are more often looking to stick with platforms (Windows, SQL , Office, etc) for longer periods of time to avoid upgrade costs and issues. Or they're looking at services like Google Apps or Office 365 to have a steady, known variable cost they can plan on per person.
We still run SQL2K8 here. No reason we can see to upgrade.
October 20, 2014 at 10:06 am
Steve Jones - SSC Editor (10/20/2014)
Gary Varga (10/20/2014)
When thinking of costing, we cannot forget what are ongoing costs, year on year, and what are repeated costs, upgrade after upgrade. If we remember this then sometimes the costs appear differently.I think we're seeing less of this. Companies are more often looking to stick with platforms (Windows, SQL , Office, etc) for longer periods of time to avoid upgrade costs and issues. Or they're looking at services like Google Apps or Office 365 to have a steady, known variable cost they can plan on per person.
We still run SQL2K8 here. No reason we can see to upgrade.
That is my point really. Even when upfront licensing costs may appear a little(?) on the high side, spread out over a decade suddenly may not seem such a hit anymore.
I know. I wasn't clear. Back to the day job. 😉
Gaz
-- Stop your grinnin' and drop your linen...they're everywhere!!!
October 21, 2014 at 7:31 am
We did a total rewrite of our internal software package. We used new technology and the project timeline ended up doubling before we were done simply because of the programmers being unfamiliar with that new technology. It was good in the end but we should have trained better beforehand or just chosen something more familiar.
October 21, 2014 at 7:42 am
Iwas Bornready (10/21/2014)
We did a total rewrite of our internal software package. We used new technology and the project timeline ended up doubling before we were done simply because of the programmers being unfamiliar with that new technology. It was good in the end but we should have trained better beforehand or just chosen something more familiar.
I always find it is better for developers to develop a throwaway prototype with new technology first. It gives the developers confidence to develop at pace and they know that a key achievement is their learning.
I tend to find that as the proper project starts with developers with at least some experience in the new technology that they progress at a faster pace without as much rework required.
Gaz
-- Stop your grinnin' and drop your linen...they're everywhere!!!
October 21, 2014 at 8:02 am
Gary Varga (10/21/2014)
Iwas Bornready (10/21/2014)
We did a total rewrite of our internal software package. We used new technology and the project timeline ended up doubling before we were done simply because of the programmers being unfamiliar with that new technology. It was good in the end but we should have trained better beforehand or just chosen something more familiar.I always find it is better for developers to develop a throwaway prototype with new technology first. It gives the developers confidence to develop at pace and they know that a key achievement is their learning.
I tend to find that as the proper project starts with developers with at least some experience in the new technology that they progress at a faster pace without as much rework required.
When working with new technology, having even one team member who has been there and done that before is priceless. The apprentice / journeyman / master model is something that the IT industry could benefit from. So often it seems we're just flying by the seat of our pants, and even years on the job amounts to practically nothing, because the experience is too narrow.
"Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho
October 21, 2014 at 8:19 am
Eric M Russell (10/21/2014)
Gary Varga (10/21/2014)
Iwas Bornready (10/21/2014)
We did a total rewrite of our internal software package. We used new technology and the project timeline ended up doubling before we were done simply because of the programmers being unfamiliar with that new technology. It was good in the end but we should have trained better beforehand or just chosen something more familiar.I always find it is better for developers to develop a throwaway prototype with new technology first. It gives the developers confidence to develop at pace and they know that a key achievement is their learning.
I tend to find that as the proper project starts with developers with at least some experience in the new technology that they progress at a faster pace without as much rework required.
When working with new technology, having even one team member who has been there and done that before is priceless. The apprentice / journeyman / master model is something that the IT industry could benefit from. So often it seems we're just flying by the seat of our pants, and even years on the job amounts to practically nothing, because the experience is too narrow.
Spot on!!!
Gaz
-- Stop your grinnin' and drop your linen...they're everywhere!!!
October 21, 2014 at 9:44 am
Ah, being both adult and wise about picking technology tools.
What fun is that? Soooo many technical religions to pick from, so little time!
The more you are prepared, the less you need it.
Viewing 15 posts - 16 through 29 (of 29 total)
You must be logged in to reply to this topic. Login to reply