November 16, 2017 at 8:22 am
Steve Jones - SSC Editor - Thursday, November 16, 2017 7:58 AMSioban Krzywicki - Wednesday, November 15, 2017 11:41 AMHi everyone,
It has been awhile since I posted here and I hope you're all doing well.
I have a weird situation and I was wondering about the opinion of the most knowledgeable database people I know, so I figured I'd go right to where you all are. : -)
It has been proposed that we replace a large, heavily relational database structure with Mongo.
What do you think? Personally I don't think it is workable. I'd love to know more arguments against attempting such a thing. Most of my googling finds older criticism and proponants claim the newest version doesn't have these problems.Want to learn Mongo? This is a good opportunity
Paid by the hour? This is the sweet smell of opportunity
Your business depends on this app/db? Update your resume.
Heh... the way I like to think of it is "Every cloud of change has a Green lining to it... for someone". 😉
--Jeff Moden
Change is inevitable... Change for the better is not.
November 16, 2017 at 8:22 am
Sioban Krzywicki - Wednesday, November 15, 2017 1:25 PMThey say it is scalability and rapid development.
I believe it is because someone heard buzzwords & they don't understand the technologies involved.
They have not done any study. They know an analytics company that does great with its 8 tables. We have 900 tables.
To make a good recommendations we'd need to know more, but, in theory MongoDB can scale better because you can just add nodes, and it can be more flexible in ingesting data because it is "schema-less", but, as one of my co-workers say, in a relational database you enforce schema going in, and in a NoSQL store you enforce it going out.
Jack Corbett
Consultant - Straight Path Solutions
Check out these links on how to get faster and more accurate answers:
Forum Etiquette: How to post data/code on a forum to get the best help
Need an Answer? Actually, No ... You Need a Question
November 16, 2017 at 8:23 am
xsevensinzx - Thursday, November 16, 2017 7:38 AMThat being said, they don't solve all problems easy. The most difficult for me in my experience is while the data is exposed to the end user, querying it in the way the data store accepts it is a challenge compared to the familarity of the traditional RDBMS. ...::sigh::
Data stores have a lot of value. Just not as the data warehouse. They are perfect for landing data before it hits the data warehouse. You cannot beat distributed processing in a sea of single machine warehousing. Have a massive query that requires a lot of indexing and optimization int he warehouse? Do it on the data store before it hits the warehouse! This is why using them also for computing that data once it lands is a great option too. But the end result, still makes sense to feed that semi-processed data into a data warehouse where it can better assist data conformity, enforce good data integrity, and of course, expose the end result in a familiar way that is accessible to all WHILE ALSO allowing you to have that raw data layer that can be exposed too.
This is how I see most of the NoSQL items. If they are used as stores, they are fast and easy for developers to work against with singletons (or low rows) of data. However, querying them is an issue. There are ways to do this, but then you've really melded an RDBMS structure on top of the store, and some of what you wanted is gone from the NoSQL side.
The MS people say, well, of course you need a DW, and you'd need that with a RDBMS, but that's not really true. Many people just run their RDBMS as a DW as well, aggregating lots of data and doing adhoc querying.
Using these NoSQL databases is nice for developers because they remove some of the object mapping they might otherwise do, and find to be a pain. However, I'm not sure they necessarily are faster in the long run. If you're starting a new app and just growing it, maybe this is the way, though you should prep a DW or plan on adding the SQL features to your store later.
November 16, 2017 at 8:53 am
Steve Jones - SSC Editor - Thursday, November 16, 2017 8:23 AMxsevensinzx - Thursday, November 16, 2017 7:38 AMThat being said, they don't solve all problems easy. The most difficult for me in my experience is while the data is exposed to the end user, querying it in the way the data store accepts it is a challenge compared to the familarity of the traditional RDBMS. ...::sigh::
Data stores have a lot of value. Just not as the data warehouse. They are perfect for landing data before it hits the data warehouse. You cannot beat distributed processing in a sea of single machine warehousing. Have a massive query that requires a lot of indexing and optimization int he warehouse? Do it on the data store before it hits the warehouse! This is why using them also for computing that data once it lands is a great option too. But the end result, still makes sense to feed that semi-processed data into a data warehouse where it can better assist data conformity, enforce good data integrity, and of course, expose the end result in a familiar way that is accessible to all WHILE ALSO allowing you to have that raw data layer that can be exposed too.
This is how I see most of the NoSQL items. If they are used as stores, they are fast and easy for developers to work against with singletons (or low rows) of data. However, querying them is an issue. There are ways to do this, but then you've really melded an RDBMS structure on top of the store, and some of what you wanted is gone from the NoSQL side.
Not just for developers though. Analyst have the benefit here as well. While I did say the RDBMS is easier for them, it doesn't stop them from at least trying. I didn't mention this in my prior statement, but I think the best value data stores have for me as the data architect is that I can quickly land data to where I can give it to my end users as an appetizer to snack on while I work on the main course (getting that data to the data warehouse). During that process of adding that new data source, they have time to be content, give feedback, and even change their mind about that main course. In a sea of users who don't always know the end result, this is killer for me because of the value it brings with trying to be a good chef and serve up the right dish.
November 17, 2017 at 7:33 am
You know it's a Friday when a general office announcement e-mail is sent out and everyone hits Reply To All to send congratulations.
Humbug.
The absence of evidence is not evidence of absence
- Martin Rees
The absence of consumable DDL, sample data and desired results is, however, evidence of the absence of my response
- Phil Parkin
November 17, 2017 at 7:54 am
Phil Parkin - Friday, November 17, 2017 7:33 AMYou know it's a Friday when a general office announcement e-mail is sent out and everyone hits Reply To All to send congratulations.
Humbug.
Uh huh. We suffer from the same affliction here. The thought process may be that if it didn't happen in an email, then it didn't really happen, so people want everyone to be spammed with know about everything. Or, there was no thought before hitting send and everyone always uses reply-all for everything, no matter how inappropriate. Like I said, we have the same plague.
November 17, 2017 at 8:18 am
Phil Parkin - Friday, November 17, 2017 7:33 AMYou know it's a Friday when a general office announcement e-mail is sent out and everyone hits Reply To All to send congratulations.
Humbug.
BCC for the win....
November 17, 2017 at 8:42 am
ZZartin - Friday, November 17, 2017 8:18 AMPhil Parkin - Friday, November 17, 2017 7:33 AMYou know it's a Friday when a general office announcement e-mail is sent out and everyone hits Reply To All to send congratulations.
Humbug.BCC for the win....
The CEO does it every Friday after lunch. Usually there's a 3-4 MB video file attached instead of the link.
Michael L John
If you assassinate a DBA, would you pull a trigger?
To properly post on a forum:
http://www.sqlservercentral.com/articles/61537/
November 17, 2017 at 8:48 am
Michael L John - Friday, November 17, 2017 8:42 AMThe CEO does it every Friday after lunch. Usually there's a 3-4 MB video file attached instead of the link.
Oh, I bet your Exchange Server LOVES that. 🙂
Thom~
Excuse my typos and sometimes awful grammar. My fingers work faster than my brain does.
Larnu.uk
November 17, 2017 at 8:52 am
Was this thread, RE: Shared stored procedure with very complicated input, removed or hidden for a reason?
November 17, 2017 at 9:01 am
Thom A - Friday, November 17, 2017 8:48 AMMichael L John - Friday, November 17, 2017 8:42 AMThe CEO does it every Friday after lunch. Usually there's a 3-4 MB video file attached instead of the link.
Oh, I bet your Exchange Server LOVES that. 🙂
No, the Exchange administrator loves it!
Michael L John
If you assassinate a DBA, would you pull a trigger?
To properly post on a forum:
http://www.sqlservercentral.com/articles/61537/
November 17, 2017 at 9:02 am
Thank you all for the input. I know I was vague, but I can't be disclosing company-sensitive stuff.
I'm glad I can come here for more opinions of people who know what they're talking about.
--------------------------------------
When you encounter a problem, if the solution isn't readily evident go back to the start and check your assumptions.
--------------------------------------
It’s unpleasantly like being drunk.
What’s so unpleasant about being drunk?
You ask a glass of water. -- Douglas Adams
November 17, 2017 at 9:34 am
Lynn Pettis - Friday, November 17, 2017 8:52 AMWas this thread, RE: Shared stored procedure with very complicated input, removed or hidden for a reason?
this one ?
________________________________________________________________
you can lead a user to data....but you cannot make them think
and remember....every day is a school day
November 17, 2017 at 9:39 am
J Livingston SQL - Friday, November 17, 2017 9:34 AMLynn Pettis - Friday, November 17, 2017 8:52 AMWas this thread, RE: Shared stored procedure with very complicated input, removed or hidden for a reason?this one ?
Yes. I received an email saying there was a new post, but clinking the link took me to a page saying it had been deleted or removed.
November 17, 2017 at 10:14 am
For your enjoyment: https://www.stavros.io/posts/startup-mistakes-datastore/
Viewing 15 posts - 60,421 through 60,435 (of 66,712 total)
You must be logged in to reply to this topic. Login to reply