This ESG Lab Report documents the validation of ClearDB’s Aventra IRON technology. ESG Lab leveraged a combination of guided demos, audited performance results, and ClearDB self-experienced quantitative cost savings to highlight the ease of deployment, as well as potential performance and cost efficiencies gained using Aventra IRON software.
As organizations continue to modernize their applications infrastructure, the goal of creating a complete, operationally efficient IT infrastructure has never been more important. Whether leveraging virtualization on-premises to consolidate and better utilize existing hardware components or moving services and applications to the cloud, organizations want to simplify existing IT processes, reduce hardware footprints, improve infrastructure performance, and complete all of it for a fraction of their current costs. And as data growth continues at a torrid pace, organizations are seeking innovative ways to get ahead of the potential burden placed on IT to deliver the service and performance levels end-users have come to expect.
Cloud adoption continues to be on an upswing, with many IT organizations sending everything they can off-premises. In fact, ESG research indicates that 36% of organizations have adopted a cloud-first policy for their newly developed and deployed applications.1 Though some are still hesitant to move to the cloud due to concerns related to control, performance, and security, the reduction in management complexity, the improved resource efficiency, and the potential reduction in costs provide sufficient benefit for organizations to look past their concerns. ESG research shows that 43% of respondents to our annual spending intentions survey selected cost reduction as one of their highest priority business outcomes over the next 12 months. The second most-cited response, selected by 40% of respondents, is increasing productivity.2 In other words, organizations are hoping to get more out of their existing resources or move to the cloud to help with cost reduction initiatives.
The Solution: Aventra IRON
Aventra IRON is performance enhancing, infrastructure optimization software for digital applications running on Linux in on-premises or cloud environments. Leveraging core Linux capabilities in combination with the unification of storage and memory, Aventra IRON enhances server compute performance to yield faster applications that allow IT to consolidate workloads from multiple applications into a virtualized, single-server instance.
Performance gains with Aventra IRON are achieved with optimization of the software stack at the storage and memory levels. For storage, since RAM and solid-state disk (SSD) storage are hundreds or thousands of times faster than spinning disk, Aventra IRON relocates large chunks of frequently used data that might otherwise be stored on spinning disk to a dedicated, unified region of storage that combines RAM with SSD, henceforth called the Aventra Storage Volume. In the best case, all data will be stored in the highly performing, low-latency Aventra Storage Volume, but for scenarios where data sets exceed that space, Aventra IRON will place the hottest data there for continuous improvement of I/O performance. With Aventra IRON installed, Linux treats the entire Aventra Storage Volume as if it were purely RAM. Data from this space is now accessed directly by the Linux kernel, eliminating the normal process latency associated with traversing the entire software stack and resulting in consistent, low-latency IOPS.
Aventra IRON provides an optimization solution for on-premises and private cloud infrastructures. Whether for a massive enterprise data center with multi-tenancy or for small co-located data centers, Aventra IRON provides applications with a general-purpose compute infrastructure that gets better use out of existing resources and delivers improved IOPS performance. In some cases, servers and related network equipment can be eliminated or repurposed. An example of the consolidation and efficiency benefits organizations can expect is shown in Figure 2.
Aventra IRON is also currently available in Amazon Web Services as a preconfigured Amazon Machine Image (AMI) and soon to be available in Azure. Organizations deploy their highly efficient Aventra IRON enabled VM in the cloud, migrate targeted workloads to it, and then shut down their standard cloud VM instances. Organizations gain the ability to either reduce the size of their existing VM instances or consolidate multiple workloads onto the same VM instance, all while improving application performance. In the best case, organizations gain both benefits, improving cloud efficiency and performance while reducing their cloud footprint and therefore cloud spend. An example of cloud consolidation with Aventra IRON is shown in Figure 3.
Aventra IRON technology can be deployed quickly and easily without disrupting existing infrastructure, applications, or their workloads. Aventra IRON runs next to the Linux operating system of the physical or virtual server; no changes are required to the Linux OS or to the application code. The only adjustment required is to point an application’s data files to the new Aventra Storage Volume.
ESG validated the simplicity of deploying an instance of Aventra. For this phase of testing, ESG focused on the deployment of Aventra IRON in Amazon Web Services (AWS). Upon logging into the AWS dashboard, the elastic compute cloud (EC2) wizard was launched. Using the AWS marketplace, four different instance types of Aventra IRON were found: small, medium, large, and extra-large. A view of the four AMIs as they appear in the marketplace is shown in Figure 4.
ESG selected the small Aventra IRON AMI, which consisted of one vCPU and two GB of RAM. After clicking through the pricing and EC2 instance types, Aventra IRON could be configured. Options related to the number of instances, Virtual Private Cloud details, and networking information were accurately auto-populated, enabling ESG to rapidly move through the wizard, which included stages related to adding storage, adding tags, and configuring security groups. Upon completion of the launch wizard, ESG was required to create a security key pair, enabling logins to the newly deployed Aventra IRON instance. The final step in the deployment was to initiate the build, which occurred by simply clicking the Launch button.
The launch of the small Aventra IRON instance completed in less than a minute, while the full system initialization, including the installation of Aventra IRON and other software, took another two to three minutes. ESG viewed the progress of the initialization through the EC2 console. To verify the successful creation of the Aventra IRON instance, ESG Lab opened a terminal session and SSHed into the instance by entering the command ssh -i ~/Desktop/esgkey.pem email@example.com
The Linux file system gains access to Aventra IRON’s unified storage as a mountable meta-device. ESG verified that the unified storage had been created and mounted during the initialization process by running a disk free command (df –h) in the open terminal. This yielded a result showing all the mounted file systems, including /dev/md127, which was mounted onto /data. This directory serves as the destination folder for all application data files that require the benefit of Aventra IRON’s technology. Figure 5 shows the terminal window with the SSH and disk free commands, as well as the results.
Why This Matters
A variety of solutions can help improve the performance of mission-critical, data-intensive applications, but most come with their own challenges. Of those solutions, the most common approach is to address performance concerns at a hardware level—faster storage, more spindles, higher powered CPUs, and more memory. The underlying commonality across these solutions is that they temporarily alleviate performance concerns at a high cost. Most software approaches either require changes to the application, the data, or in the worst case, both. This leads to lengthy, complex deployments with higher risk of downtime or data loss. With the growing size of application datasets and diversity of application workloads, organizations are thinking outside of the traditional paradigms to meet their performance and scalability requirements while meeting their strict budget requirements.
ESG Lab validated the speed and simplicity of deploying Aventra IRON on Amazon Web Services EC2. This is especially true for IT administrators with existing knowledge of AWS environments and the AWS marketplace. In less than ten minutes, ESG selected, configured, and deployed a new small instance of Aventra IRON. And since Aventra IRON does not require changes to existing applications or their data, as soon as everything is migrated to the new Aventra IRON instance, storage-based application performance bottlenecks are alleviated. Furthermore, the ability to consolidate by running multiple applications within a single Aventra IRON instance helps organizations reduce their cloud footprint and costs.
Performance and Cost Efficiency
Aventra IRON reduces the requirement to traverse the software stack to satisfy I/O requests, enabling enhanced performance when compared with a traditional, bare-metal infrastructure. ESG Lab audited performance results run by the ClearDB team to highlight the performance benefits Aventra IRON can yield in a production OLTP environment.
Testing was completed using Sysbench, an industry-standard benchmarking tool that can simulate real-world MySQL database environments and workloads. Within the tool, real databases are built and then exercised based on provided test profiles or customized workloads. This enables results that not only include traditional storage performance metrics such as read and write IOPS, but database-specific metrics such as transactions/sec and average transaction latency, which are often more meaningful to database administrators.
A Sysbench-provided test profile that simulated an OLTP environment was used for testing. It consisted of common OLTP database queries, including selects, updates, deletes, and inserts that yielded an average I/O ratio of 75% reads and 25% writes. The test has a defined number of queries that must be executed. Three different configurations were tested, all of which were configured on the same x86 server consisting of a single 4-core processor running at 2.4GHz with 64 GB of RAM and a 1TB PCIe flash device. The three configurations included a standalone instance of MySQL with full two-phase commits enabled, a standard instance of MySQL with asynchronous two-phase commits enabled, and MySQL with Aventra IRON.
Three database deployments were created from scratch on each configuration:
- 100 tables with 1,000 rows per table.
- 10 tables with 1,000,000 rows per table.
- 200 tables with 100,000 rows per table.
Performance results were compared across the three configurations and ESG analyzed the transactions/sec and average latency. As shown in Figure 6, Aventra IRON far outperformed all but one of the standalone MySQL deployments in terms of transactions/sec. Table 1 shows the transactions/sec and corresponding read/write requests/sec as reported by Sysbench. Across all the tests, Aventra IRON yielded an average performance gain of 3.2X compared with the MySQL standalone instances.
Latency is just as important, if not more so, in OLTP environments as it directly impacts the end-user experience. With each test set to complete approximately 10,000 total events, ESG also viewed the total time to complete each test. In theory, the faster the performance, the lower the latency and the less time required to complete the test. As expected, Aventra IRON produced the lowest average latency across the three tests (see Figure 7). The average latency for the standalone MySQL instance averaged approximately 86 ms across the three tests. With async, that average was reduced to 55 ms. With Aventra IRON, average latency across the three tests averaged under 30 ms, which represented a nearly 70% reduction in average latency. The average reduction in latency allowed Aventra IRON to complete the tests in no more than 6 seconds, while standalone MySQL and MySQL with async took averages of 13.5 and 8.5 seconds across tests.
After validating performance benefits offered by ClearDB, it was clear to ESG that the gains in performance did not just enable existing applications to perform better, but it also enabled organizations to consolidate their application infrastructure. For example, with an ESG-witnessed performance gain of as much as 4.5X for a single application using Aventra IRON, in theory, an organization could deploy multiple applications to consume that available performance, especially if a single application does not require that much performance. The potential benefits are substantial. Imagine those four applications each residing on its own physical server. By consolidating them onto a single server without impacting performance and availability, organizations not only save on capital expenditures (three extra servers), but they also improve existing resource efficiency by getting more out of the single server remaining.
ClearDB put its technology to the test and deployed it within its own infrastructure. ESG was given access to ClearDB’s internal performance monitoring and cost data before, during, and after the transition from a traditional infrastructure without Aventra IRON to an efficient, performance-optimized infrastructure with Aventra IRON.
First, ESG looked at the performance benefits ClearDB achieved in the infrastructure, which consisted of a mix of physical and virtual resources deployed in the traditional siloed approach. Figure 8 highlights the reads/sec and writes/sec reported by ClearDB’s IT administrators over a 16-month period. After seeing performance begin to bottleneck, IT had the choice to either purchase more hardware or deploy Aventra IRON. The benefits are obvious: ClearDB achieved more than 2X the performance without major changes to its hardware infrastructure by better utilizing existing hardware and thus, extended the benefit of traditional virtualization. Further, the maximum supported reads/sec increased by 2.8X, while the writes/sec increased an impressive 10X.
ESG also looked at the cost savings, increase in supported capacity (number of unique databases) associated with deploying Aventra IRON in the cloud, and utilization. We analyzed data from before, during, and after ClearDB’s infrastructure transition to Aventra IRON. Table 2 highlights the cost and capacity differences of a default AWS instance and an Aventra IRON instance as reported by ClearDB. The monthly cost of a default node in AWS was $1,900 while supporting a capacity of 4,000 small, unique databases per node. An Aventra IRON AWS instance cost 60% more, but is quickly offset by the 5X increase in available capacity of 20,000. Table 2 highlights these differences, including the 69% savings in cost/capacity.
To understand the cost and consolidation efforts put forward by ClearDB to transition to an all Aventra IRON cloud cluster, ESG analyzed the ongoing monthly cluster costs and consumed capacity before, during, and after the transition in Table 3.
ClearDB started with a cluster consisting of 27 default AWS nodes that cost $51,300 per month. Of the available capacity, 69,218 or 64% of it was consumed, yielding a cost/utilized capacity of $0.74. During the transition, four Aventra IRON AWS instances were deployed, enabling the reduction of 20 default AWS instances. The remaining seven AWS instances cost a total of $13,300 and 77% of their available capacity was consumed (21,562), while the four new Aventra IRON instances cost 10% less at $12,000 and 60% of their available capacity was consumed (47,656). The cost/capacity savings offered by Aventra IRON was nearly 60%. The final transition point was to take the remaining default AWS instances and migrate their workloads to the four Aventra IRON instances. Though the total available capacity on the four Aventra IRON instances (80,000) was less than the starting point (108,000), the four-node Aventra IRON cluster was still considered underutilized at just 60% of capacity being consumed. This enabled the full transition of the 69,218 capacity to the four-node Aventra IRON cluster. The result of the full transition to an Aventra IRON-only cluster was ClearDB better utilizing less resources to save $39,300 per month, which proved to be a 77% reduction in monthly costs, while continuing to deliver scalable performance.
Again, it should be noted that the measurement of database capacity in ClearDB’s case was defined as the total number of databases. This particular definition of database capacity may not be applicable to everyone, but that should not be a deterrent to organizations with other requirements or definitions. Some view database capacity to mean the number of supported customers within a single instance, or even more specifically, the number of concurrent users accessing the database at a given time. In fact, one organization that turned to ClearDB had a requirement to increase the number of concurrent authenticated and registered connections by 2X. Upon switching to ClearDB, they increased their concurrent user count from 5,000 to 10,000 without needing to change their applications or a majority of their infrastructure.
Why This Matters
Though the cost to maintain an already deployed IT infrastructure may appear low, the cost to scale can prove to be massive. To meet the scalability requirements of a rapidly growing business, adding new servers and storage makes an already hard-to-manage infrastructure even more complex. The adoption of virtualization helps organizations avoid these large upfront hardware costs by consolidating existing resources, but performance concerns limit organizations from fully transitioning some performance-intensive applications such as mission-critical databases. Organizations are looking for ways to meet their scalability requirements while minimizing cost and complexity by reducing new capital expenditures and getting the most out of their existing, underutilized resources. One of those ways is moving the infrastructure closer to the applications, especially applications with critical data that remains resident. In these cases, memory serves as an ideal place to handle requests, allowing the workload to be serviced closer to the application rather than traversing the entire infrastructure and software stack.
ESG Lab validated the potential performance and cost efficiencies that can be achieved by leveraging Aventra IRON on-premises. Audited performance results and ClearDB’s own experience from deploying Aventra IRON in their own infrastructure confirmed impressive performance gains and cost reductions. A real-world OLTP workload with a boost from Aventra IRON yielded an average performance gain of 3.2X transactions/sec and a latency reduction of nearly 70% when compared with a standalone MySQL database deployment running the same workload. Further, when ClearDB deployed Aventra IRON in its own infrastructure, storage I/O performance more than doubled. Not only did ClearDB reduce their monthly costs by 77% after switching to Aventra IRON, but they also increased their available capacity per node while continuing to meet their performance SLAs at scale.
The Bigger Truth
Organizations are rapidly adopting technologies that help improve IT infrastructure efficiency and reduce costs. For on-premises or private cloud infrastructures, virtualization helps organizations get more from existing resources, reducing costs. However, performance concerns and potential complexity when needing to scale serve as a deterrent. Shifting to a public cloud environment can help meet scalability, elasticity, and simplicity requirements, but performance, control, and security serve as a deterrent. Further, the ongoing costs of leveraging the cloud may be more than expected. With the continuous growth of data and the diversity of application workloads that need to be serviced by a given infrastructure, traditional paradigms for addressing such needs no longer suffice.
The Aventra IRON technology enables organizations to get more out of their existing resources while continuing to meet strict performance SLAs. By combining the consolidation capabilities of virtualization with the performance capabilities of memory and flash storage, organizations gain technology that unifies memory and storage to deliver a highly efficient software stack that commonly exceeds the needs of existing application workloads. By minimizing the distance that a majority of I/O requests need to travel to be serviced, the density on existing server resources is greatly increased. This means organizations get more out of what they already paid for while reducing the need to purchase more hardware to meet the increasing capacity and performance demands of their mission-critical applications. And since the technology is infrastructure-, application-, database-, and cache-independent, implementing the solution in an existing environment takes only minutes.
Whether on-premises or in the cloud, Aventra IRON can reduce your infrastructure footprint without sacrificing performance. In fact, ESG witnessed impressive performance improvements when using Aventra IRON in an emulated real-world OLTP environment—4.5X more transactions/sec and nearly 70% lower average latency. These types of benefits led ClearDB to deploy the technology in its own global database-as-a-service (DBaaS) environment, yielding impressive efficiency and cost savings. ClearDB reduced their cloud footprint by 85% and therefore reduced their monthly costs by 77%, all while continuing to meet their performance and scalability requirements.
With Aventra IRON, organizations gain improved infrastructure and application efficiency while reducing their bottom line. Simply install the software, migrate existing application data to the proper directory, and let Aventra IRON take care of the rest.
1. Source: ESG Research Report, 2017 Public Cloud Computing Trends, April 2017.↩
2. Source: ESG Research Report, 2017 IT Spending Intentions Survey, March 2017.↩