Voice of the DBA

Tag: NoSQL

  • How Much Eventual Consistency Do You Have?

    There’s a never ending debate about NoSQL v RDBMS systems that seems to polarize the people that prefer one technology over the other. In fact, it seems that every time I talk to someone that dislikes NoSQL technology in general, the topic of eventual consistency comes up. The idea that not all data in our system might be up to date seems to be one of the concepts that scares many DBAs most of all.

    I was chatting with some people recently about complex SQL Server configurations, and the topic of replication came up. While this is a technology with so much potential, it seems that replication has been a bit neglected by Microsoft and is both amazing and brittle in its implementation. However, if you think about it, replication results in data that isn’t consistent across the systems.

    I’m sure many would argue that this isn’t an issue, but how many of you have businesses that make decisions or have processes built on data in replicated databases? I’m sure plenty of you, and most of the time, the data is consistent enough for use by our clients.

    There are plenty of ways in which we implement data movement across our databases that results in potentially inconsistent results for our clients. In fact, I’ve had no shortage of discussions with clients that can’t understand why two reports run minutes apart show different results.

    Today I’m curious how many of you have systems that your businesses depend on where the data is eventually consistent because of some technology that moves information from one database to another. Perhaps you might even share some of the tricks you use to ensure that delays or problems in your transfer process are detected and fixed before your clients realize just how inconsistent their data might be.

    Steve Jones

    The Voice of the DBA Podcast

    Listen to the MP3 Audio ( 2.8MB) podcast or subscribe to the feed at iTunes and LibSyn.

  • NoSQL Complaints

    I think that NoSQL databases are a really interesting set of technologies, however they aren’t necessarily a panacea for many problems, including scalability. There are likely some problem domains where a particular NoSQL technology might work better than an RDBMS, and perhaps some situations where many types of databases would work well.

    I ran across a piece that looked at some of the complaints that companies have with NoSQL technologies. It’s a generic piece, perhaps too generic, since a graph database is not like a document database. However it does talk about the issues that companies see, such as performing analytic queries. While a row-based RDBMS isn’t ideal for these queries, often the platforms work very well for a wide variety of reporting requirements.

    I do think some of the problems people have noted have to do with the maturity of the systems. There are a lack of references for how to run a high uptime NoSQL database, not many data models to review that may or may not work well, and relatively few administrative techniques to maintain these databases. The problem isn’t that NoSQL databases can’t meet your needs, or run quickly, or provide many 9s of uptime, but rather many enterprise professionals might not know how to find the information to accomplish these goals. That’s something easily solved across time.

    Ultimately I think the NoSQL world has a lot of maturing to do, and we, as professionals and media, need to start to differentiate among the various flavors of databases. I am glad to start seeing articles and blogs refer to graph databases when they cover a product like Neo4j rather than writing NoSQL. Slowly I am seeing MongoDB referred to more often as a document database, not a NoSQL database. It’s good to start to separate technologies and not confuse one with another.

    One thing in the piece I did notice is that it’s likely that we will see more platforms and products incorporate both RDBMS and non-RDBMS technologies together. We see that in SQL Server already with the addition of columnstore technologies in the last few versions and Polybase to SQL Server 2016. I expect we’ll also start to see some of the capabilities of DocumentDB added to SQL Server, along with, perhaps, graph based technologies in the future.

    Steve Jones

    The Voice of the DBA Podcast

    Listen to the MP3 Audio ( 2.7MB) podcast or subscribe to the feed at iTunes and LibSyn.

  • Relational By Default

    I’ve read a few pieces from Vladimir Khorikov and I tend to like the thoughtful way in which he approaches building software. When I saw a link to his SQL v NoSQL databases, I was intrigued to see what his thoughts would be. It’s a good read, though with relatively few comments or debates posted at the time I read the article. I was hoping for more and I encourage you to add your thoughts.

    The main contentions in the piece are that any of the simpler development techniques that we can use with NoSQL databases don’t remove the need for implementing data quality or management features. They just require those features to be implemented by the developer. Specifically the need to deal with multiple schema versions over time, or the requirement we enforce parent child relationships.

    Perhaps that’s not too bad. After all, if we evolve schemas and the code can easily deal with the changes, that’s good. However I think the point about having relational storage when you can, and adding in other stores makes sense. This is what I really believe as well, since we often need to query and combine data, which is something relational stores do very well and very efficiently.

    Ultimately I’m sure we will continue to see arguments about relational and NoSQL models being better or worse for particular problems. I actually welcome the arguments  because there are certainly domains of problems where one system works better than the other, and it’s good to develop some clarity about those particular problems. However there are also likely many situations where multiple platforms can work, and in those cases, we should use what works well for our team.

    Steve Jones

    The Voice of the DBA Podcast

    Listen to the MP3 Audio ( 1.9MB) podcast or subscribe to the feed at iTunes and LibSyn.

  • NoSQL Inside SQL Server

    There has been quite a debate the last few years between using NoSQL and SQL Server as data stores for different applications. We’ve seen a great deal of interest from developers in document stores as the data structures are closer to what a C# or Java developer sees in their code. Graph databases are becoming popular for certain types of data as well, since their query performance can be extremely powerful in some cases. There are also the scalability aspects of NoSQL systems touted as a better way to build Internet based applications.

    However NoSQL databases aren’t the ultimate data store. There are pros and cons to using them, and while it might be easier to being development on one for an application developer, I would argue it’s just as easy to get caught mis-using a graph database or a document database as it is to poorly program a relational database.
    This week, in advance of the SQL Server 2016 release, I caught a post from the SQL Server storage engine team showing how JSON storage inside of a table can provide some NoSQL-like benefits to applications. It’s a good idea, but like many NoSQL solutions, you need to spend a bit of time examining your particular requirements. If you need to update your JSON often, especially in large batches, this might not prove to be a good idea, especially if you need consistency.
    I think ultimately that everything does relate to the CAP theorem, precisely because data has to exist somewhere, and anytime you need to duplicate it, move it, aggregate it, or anything to relieve some other issue, then there isn’t any magic taking place. You’re going to have to do work to move the data. Some platforms may make that easier, but I bet there are other drawbacks that you will have to consider.
    Ultimately I think you need to pick the platform that best suits your environments. That’s not only the application requirements, but also the developers’ skills, the operational staff’s ability to manage the platform, and the maturity of the platform. Those are tough questions, and while I’m glad people experiment with new systems, I do hope that they think carefully about the positives and negatives, and more importantly, share their experiences with the rest of our industry.

    Steve Jones