Top 3 causes of storage bottlenecks

Storage bottlenecks can affect application performance—and cause you to lose customers. It’s essential, therefore, to understand and resolve them.

What are the leading causes of storage bottlenecks? They include mismanagement of virtual storage, applications with insufficient or the wrong type of storage, and poor storage design. Let’s explore these further.

1.    Mismanagement of virtual storage

If you’re not monitoring your IT infrastructure, carving up your storage array or subsystem between numerous virtual machines (VMs) can be challenging. That’s because you have no visibility to the flow of traffic. Allocating your VMs to a logical unit number (LUN) without data to guide you is like building a highway without doing a traffic study to assess traffic volume at different times of the day and week. How would you know how many lanes are required to accommodate the traffic?

+ Also on Network World: Bandwidth alone won’t solve application performance problems +

Alternatively, let’s say the four-lane highway already exists, as does your terabyte of storage. Without studies, the traffic engineer cannot know how many vehicles are on the road and when they’re there. Similarly, without monitoring, you wouldn’t be able to measure your virtual guests’ demands on storage at all times.

It’s possible that the activities of your 10 busiest VMs could be restricted because they’re all on the same LUN. Just like traffic around Los Angeles slows to a crawl (or worse) in peak traffic hours, as data transactions mount, response time increases.

Part of what makes this so complicated is that it’s not the number of VMs that make the difference. It’s the nature of them. Some are actively moving data, like vehicles going to and from work on the highway every day. Others, like Los Angeles’ residents who take weekend jaunts to Santa Barbara, have only occasional needs. The challenge is pairing the most active VMs with the least active to balance out demands on storage.

2.    Applications with inadequate storage

If you have an I/O-heavy application that’s not placed correctly within a storage array, the user experience is at risk. For example, if a web application is running on the wrong type of storage or it doesn’t have enough storage devoted to it, a customer can suffer delays when trying to buy a shirt. And when the web page takes too long to populate, the customer may click away. 

3.    Flawed storage design

Today, several tiers of storage are available—tier zero, one and two. Tier zero is the fastest, one is middle of the road, and two moves like molasses. Sometimes these tiers are incorrectly managed, leading to problems.

For instance, a web app may be on tier one when it should be on tier zero. When a customer hits a query, saying, "Hey, show me what this shirt looks like in blue," the application is slow to fetch the image, and the customer is off to find the competition.

Another storage design issue is accessibility—how the servers are connected to the storage via the SAN fabric. Perhaps the server can’t get to the storage fast enough because it has to go through three connections instead of one.

To solve your storage bottlenecks, you need to monitor the performance of LUNs, applications and servers. Monitoring can help you to design efficient and effective storage, as well as anticipate and adjust to traffic volumes.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2016 IDG Communications, Inc.