Scaling Database Applications in the Cloud [White Paper]

Cloud service providers (CSPs) have long touted the substantial economies of scale that are available in the cloud to both software developers and Software-as-a-Service (SaaS) providers. However, achieving scalability and performance while preserving the cloud’s cost-saving advantages can be challenging in a multi-tenant environment. The most difficult applications to optimize for peak performance and cost savings are those that utilize a database. The reason for this is often due to database licensing arrangements. Many of these arrangements often make high-performance configurations prohibitively expensive.

Our latest white paper, Scaling Database Applications in the Cloud for Peak Price and Performance, can help software and SaaS providers maximize price and performance, in one of the most challenging IT environments – a database application for a multi-tenant SaaS offering running in the cloud. Read on for some key points from the white paper.

Scaling Up Storage I/O

Developers that want to scale performance in a cloud infrastructure typically have two options:

  • Scale out performance – this involves adding more processing power by adding more virtual machines (VMs), more processor sockets or cores, and/or more servers. A major drawback to this approach is that most applications need additional software licenses.
  • Scaling up performance – this option may or may not involved additional licensing fees. There are many methods in which to scale out including: scaling up application performance, adding more memory for caching, and using direct attached storage. Yet all of these all offer minimal improvement with some drawbacks. The best solution? Scaling up storage I/O (input/output).

The design and sophistication of storage devices have evolved over the past 50 years, with sold state drives (SSDs) and all-flash solutions emerging as the current, preferred solution. A key capability of these solutions is support for Quality of Service (QoS) – assuring performance in a multi-tenant environment where “noisy neighbors” can undermine performance.

According to Henry Baltazar, Senior Analyst of Infrastructure and Operations at Forrester Research, “‘Noisy neighbors’ disrupt cloud and enterprise storage delivery . . . This limitation is a major reason why it is so difficult for cloud storage providers to create consistent multi-tenant environments using traditional storage systems which lack Storage QoS.” This is a key reason why Baltazar calls storage QoS a “must-have” capability in the multi-tenant cloud.

Storage QoS and the cloud

The best storage QoS solutions go beyond simple rate-limiting and “best-effort” prioritization mechanisms to ensure that each application gets the performance its needs at all times under all circumstances. The only way to ensure that business-critical applications always get the performance they require is to take a Min/Max/Burst approach to managing IOPS:

  • Min IOPS is needed to ensure every volume gets a guaranteed minimum level of performance, regardless of system conditions or application activity.
  • Max IOPS is needed to set a maximum level of performance that each volume receives over time as a means to ensure fairness among all volumes.
  • Burst IOPS is needed to accommodate the occasional spikes in demand that occur in virtually all database applications.

Scaling up database application performance 

While many CSPs offer similar scale out and scale up options, few can accommodate the demands of a SaaS provider with a multi-tenant database. SaaS providers looking to engage with a CSP to handle a multi-tenant environment need to go beyond the usual criteria and work with one that offers the following:

  • Storage QoS to ensure that the IOPS performance gains will be predictable and consistent over time under changing configurations and conditions
  • A development/test environment to characterize solid state drive (SSD) versus hard disk drives (HDD) performance under varying loads and circumstances
  • Storage encryption that does not compromise performance to help ensure appropriate security controls over security or privacy
  • Support for hybrid architectures that combine the best of public clouds for cost-effective scalability, private clouds for security, and dedicated servers and/or colocation for special needs (including database clusters)

All-flash Storage from HOSTING

HOSTING’s SolidFire storage solution with QoS gives customers the ability to guarantee IOPS for their applications on a 100% solid state drive platform. Performance and capability can be scaled dynamically, as needed and without disruption to the application. Performance with the  all-flash storage array increases four-fold when compared to traditional mechanical drive based arrays. This dramatic increase in performance is what enables HOSTING to offer an always-on, QoS-backed service level agreement for mission-critical applications.

Download our complimentary white paper to learn how you can attain peak price and performance for your database applications through the HOSTING All-flash Storage Solution. You can speak to our database experts anytime to discuss your specific needs.

Scaling Database Applications In The Cloud For Peak Price And Performance

Leave a Reply

Your email address will not be published. Required fields are marked *