VM Storage Performance in the Cloud

Joey D’Antoni explains how storage architecture has changed from on-prem to the cloud:

This architecture design dates back to when a storage LUN was literally a built of a few disks, and we wanted to ensure that there were enough I/O operations per second to service the needs of the SQL Server, because we only had the available IO of a few disks.

As virtualization became popular storage architectures changes and the a SAN lun was carved out into many small extents (typically 512k-1MB depending on vendor) across the entire array. What this meant was that with modern storage there was no need to separate logs and data files, however some DBAs did, however in an on-premises world there was no penalty for this.

It’s important to keep up on these changes.

Machine Learning and Delta Lake

Brenner Heintz and Denny Lee walk us through solving data engineering problems with Delta Lake:

As a result, companies tend to have a lot of raw, unstructured data that they’ve collected from various sources sitting stagnant in data lakes. Without a way to reliably combine historical data with real-time streaming data, and add structure to the data so that it can be fed into machine learning models, these data lakes can quickly become convoluted, unorganized messes that have given rise to the term “data swamps.”

Before a single data point has been transformed or analyzed, data engineers have already run into their first dilemma: how to bring together processing of historical (“batch”) data, and real-time streaming data. Traditionally, one might use a lambda architecture to bridge this gap, but that presents problems of its own stemming from lambda’s complexity, as well as its tendency to cause data loss or corruption.

Read the whole thing.

Hooking SQL Server to Kafka

Niels Berglund has an interesting scenario for us:

We see how the procedure in Code Snippet 2 takes relevant gameplay details and inserts them into the dbo.tb_GamePlay table.

In our scenario, we want to stream the individual gameplay events, but we cannot alter the services which generate the gameplay. We instead decide to generate the event from the database using, as we mentioned above, the SQL Server Extensibility Framework.

Click through for the scenario in depth and how to use Java to tie together SQL Server and Kafka.

Choosing Clustered Index Columns

Ed Elliott wades into the clustered index debate:

I have seen this debated in forums spread over the internet for decades, and the advice that we gave ten years ago isn’t as valid today as it was then. Ten years ago, memory was considerably less, and disks were spinning rust. The advent of SSD’s and the ability to get servers with more memory than data, even on large systems have changed how we should think about designing and maintaining databases.

I generally subscribe to the NUSE philosophy: Narrow, Unique, Static, Ever-Increasing. That generally leads me to selecting identity integers or longs. For junction tables (whose entire purpose is to join two tables together and which never get referenced outside of that), I use the primary key as the clustered index.

In extreme insert scenarios, I can see wanting to maximize fragmentation in order to insert into more pages in the B-tree and avoid hot spot pages.

Event-Driven Microservices

Saeed Barghi gives us an overview of what event-driven microservices are:

Modern Microservices are all about making systems event-driven: instead of making remote requests and waiting for the response (services and components calling each other and tell each other what to do), we can send notifications to related microservices when an event occurs.

These events are facts about the business. For example, an ATM or online transaction, a new log entry, or a customer registering for a new mobile plan. They are the data points collected by organizations that make their datasets. The good thing is, we can store these events in the very same infrastructure that we use to broadcast them: Apache Kafka. The better thing is we can even process them in the same infrastructure with Stream Processing applications. This means our applications and systems are linked via this central data pipeline, that is capable of real time data broadcast and processing and all data sources are shared via this data pipeline.

Read the whole thing.

Finding Queries to Cache In-App

Brent Ozar provides guidance on the types of queries you might want to cache in your application:

Question 2: Will out-of-date data really hurt? Some data absolutely, positively has to be up to the millisecond, but you’d be surprised how often I see data frequently queried out of the database when the freshness doesn’t really matter. For example, one client discovered they were fetching live inventory data on every product page load, but the inventory number didn’t even matter at that time. During sales, the only inventory count that mattered was when the customer completed checkout – inventory was checked then, and orders only allowed if products were still in inventory then.

There are some good questions in here which can help figure out what can fit in a cache and what really needs to be fresh.

Problems Distributed Systems Experience

RJ Zaworski gives us examples of the types of problems you can run into with distributed systems:

Time limits: ending the neverending
Here’s one to ponder: how long can a long-running action go on before the customer (even a very patient, very digital customer) loses all interest in the outcome?
Pull up a chair. With no upper bound, we could be here a while.

Read on for more in that vein with JavaScript-y solutions.

Finding Dependency Clusters

Michael J. Swart performs cluster analysis with tables:

That ball of mush in the middle is hard to look at, but the smaller disconnected bits aren’t! Just like Ben, I want to work on those smaller pieces too! And just like the lonely tables we looked at last week, these small isolated components are also good candidates for extracting from SQL Server.

The script looks at joins in execution plans, which is a rather clever way of doing this when you don’t have a comprehensive set of foreign key constraints.

Big Data Often Isn’t

Arnon Rotem-gal-oz argues that “big data” is often a misnomer:

I couldn’t find numbers from Google but others say that by 2017 Google processed over 20PB a day (not to mention answering 40K search queries/second) so Google is definitely in the big data game. The numbers go down fast after that, even for companies who are really big data companies — Facebook presented back in 2017 that they handle 500TB+ of new data daily, the whole of Twitter’s data as of May 2018 was around 300PB, and Uber reported their data warehouse is in the 100+ PB range.

Ok, but what about the rest of us? Let’s take a look at an example.

I often fight with this myself—SQL Server can easily handle multi-billion row data sets, for example. It’s the same problem in Azure with SQL Data Warehouse: the “you must be this tall to ride the rides” marker is set pretty high.

Building the Right Architecture for the Job

Gogula Aryalingam takes us through an example where the newest and most expensive tools aren’t the best for the job at hand:

When Azure SQL Data Warehouse was chosen to implement a multi-dimensional data warehouse, it may have seemed like the ideal choice. Why? because it was plain to see: keywords: “SQL”, “Warehouse”. However, no, SQL Data Warehouse is ideal only when you have data loads that are quite high, not when it is only several 100GBs. Armed with a few more reasons as to why not (A good reference for choosing Azure SQL Data Warehouse), I had confronted them. But the rebuke then was that they did get good enough performance, and that cost wasn’t a problem. Until of course a few months later when complex queries started hitting the system, and despite being able to afford that cost, the value of paying that amount did not seem worth it.

Having a good architectural understanding of the Azure or AWS platform—even if you aren’t deeply familiar with all of the tools—can help avoid these types of problems.

Categories

September 2019
MTWTFSS
« Aug  
 1
2345678
9101112131415
16171819202122
23242526272829
30