“At this time, Wikibon cannot recommend running large-scale Oracle Database mission-critical workloads... on AWS ”
- Wikibon Research Jan 2021
According to Gartner, the total Database Management System (DBMS - Oracle, SQL, MySQL, PostgreSQL, etc) market hit $80B in 2021, up 22.3% from the previous year. Of this total, 49% has already shifted to the cloud, and it is predicted to grow at an astounding 53.6% CAGR until 2026, according to the 360 Research Reports.
The primary reason for the rapid increase in cloud databases is that organizations are increasingly attempting to move everything to the cloud. While databases tend to be the biggest obstacle to achieving this objective, successfully running databases in the cloud increases scalability, enhances reliability, and reduces, in some circumstances, costs. Let's see how this is achieved:
- Scalability: The cloud platform eliminates the need to purchase and stand up new hardware, which enables rapid scalability and in turn accelerates business agility, and faster time-to-market.
- Reliability: The top cloud providers offer great uptime capabilities that minimize or eliminate disruptions.
- Resource Accessibility: The ability to draw on cloud resources such as AI or big data capabilities and to always run up-to-date on hardware and operating systems makes resources easily accessible.
- Reduce (some) costs: Zero hardware costs or costs of administering the host machine operating system lead to a possible reduction or elimination of data center costs.
- Security: The public cloud is typically as, or more, secure than traditional on-premises data centers.
The cloud, however, is not a panacea; it has its challenges. Additionally, moving to the cloud itself entails a big decision - Should a company self-manage its databases or rely on a cloud management service such as AWS RDS or Azure SQL Database?
The Problem with Running Databases on Cloud Today
According to the Forrester State of Public Cloud Migration (2022) report, the second-biggest challenge (that amounts to 38%) of migrating to the public cloud is “performance issues with workloads migrated to the cloud.” The cloud infrastructure built for general-purpose workloads has been repurposed for databases that demand consistent high performance. To deliver high performance at scale, the clouds have introduced an expensive meter called provisioned IOPS and capped it (with very few exceptions) at 80K IOPS. This is a relatively low upper limit on the IOPS when compared with on-premises infrastructure, which makes today's cloud DB infrastructure extremely sensitive to workload IOPS requirements.
As mentioned in the Forester survey and other reputed surveys, security is the biggest concern about migrating to the cloud. Though the cloud providers maintain dedicated security experts focusing on deploying world-class security architecture, this benefit is negated when it comes to databases because of the inability to easily and securely share them. The only way an administrator can share a database is by giving the user access to the database account, including the credentials. This creates an obvious vulnerability.
Another challenge with databases in the cloud (and one not mitigated by managed database services) is the lack of hierarchical ownership for both databases and compute instances. For example, if the DBA gives someone the authority to create VMs but wants to limit them to 32 vCPUs, the DBA has no way to verify if the person is following the directive. Similarly, if the DBA gives someone the authority to duplicate databases, that person can only clone a snapshot of the DBA's existing databases. This limitation can create security concerns while also meaning that there is no simple way to share databases for developers, testing teams, Quality Analysts, UAT, etc.
The Pros of Existing Managed Database Services
Both AWS RDS and Azure SQL Database offer significant advantages, especially for monitoring and patching. They reduce the need for DBAs and SREs to configure and monitor databases or create snapshots, security patches, and updates. This, in turn, reduces the probability of human error.
Scaling a self-managed database in the cloud requires a sequence of steps that the DBA needs to perform manually. RDS or Azure SQL Database goes through these steps in the backend, whereas the customer experiences scalability just with the click of a button.
The Cons of Existing Managed Database Services
The first big disadvantage of RDS or Azure SQL Database is the cost. In addition to the IOPS metering for self-managed databases, assume a 70% - 100% price premium for the managed service. The managed database services do little to mitigate the limitations of cloud database performance. In February 2022, Oracle produced a video poking fun at running databases on AWS RDS.
Another big limitation with existing managed database services is the inability to create database backups. These services provide storage snapshots, which are essentially a picture of a disk. Unlike a backup that is universally known regardless of where it was created, a storage snapshot can only be understood by those who created the snapshot. This means that the data isn't yours; it belongs to the cloud providers, which makes it taxing to move data back to your on-premises environment or a different cloud. The complete reliance on snapshots entails additional security issues because of the need to grant users access to snapshots to share data.
From an administrative perspective, managed database services take away control, which is the exact opposite of how DBAs are accustomed to managing databases on-premises. On the cloud, DBAs can access their data but can't take backups - all they see is metadata of the snapshots. They can't control the VMs or where the data resides, which means that they can't comply with compliance requirements to demonstrate VM/database high availability failover.
Finally, the managed DB services do not offer anything in the way of differentiated data management. If DBAs want to sanitize data before providing it to stakeholders, they need to integrate with third-party services. There are limited dashboards for monitoring both cost and performance. Also, there is no way to curate the cloud to resolve the lack of hierarchical ownership for database and compute instances.
How Tessell is Disrupting Database As a Service
Enter Tessell to the rescue!
Tessell is a managed database service that utilizes breakthrough technology to enable customers to both reduce costs and enhance business agility. Tessell has a patent-pending technology that provides the performance benefits of NVMe storage on AWS and Azure. Tessell's technology innovation is of the same magnitude as Citrix virtualizing Windows Server, VMware virtualizing compute, and Snowflake separating compute from storage. Tessell's underpinning technology advancement gives it a competitive edge on the performance and price factors.
Competitive Performance and Cost
Tessell provides up to 2 million IOPS and eliminates all cloud metering. Even for small instance shapes and low IOPS requirements, Tessell easily beats AWS RDS and Azure SQL Database in terms of cost. As IOPS demands increase, this delta becomes much more significant. Tessell offers 10X the performance and at one-half the cost of the two current leaders. Going to reserved instances makes the delta even starker.
Governance @ your terms
Tessell allows DBaaS @ your terms. Tessell gives you the option of bringing your own existing cloud infrastructure (networks, security policies, and the cloud account itself). You are not locked into a specific data cloud but can move between clouds. You are not even locked into Tessell - you can download backups (something that AWS or Azure don't allow).
Cloud Curation
Tessell allows granular cloud management so organizations can further optimize for performance while minimizing costs. You can perform a variety of curations. Here are a few examples:
- Limit the size of instance shapes that developers can provision with a click of a button
- Automatically sanitize and limit the data passed on to the stakeholders
- Standardize database deployments by customizing database parameters
- Customize SLAs per compliance needs
- Minimize license requirements by restricting subscriptions to only Production or QA at a time or specify data locality to enable European employees only for creating or accessing databases in EMEA for GDPR compliance and other purposes.
Data Management Automation
Just like any consumer-grade application, Tessell understands, optimizes, reports, and forecasts your database costs based on a utility-like consumption model. Tessell also introduces a new concept of utilizing data apps (both Tessell and soon 3rd party apps) to effect a level of management not offered in any other database realm. These apps provide rich reporting and in-depth performance and financial analytics. They allow the automatic sanitizing of data and sharing with different user groups such as Dev, QA, UAT, etc.
Tessell Makes Life Easier
Tessell makes life both easier and better. It incorporates not only all the advantages of AWS RDS and Azure SQL but also provides hourly compute metering by letting you pay only for what you use. You can either pay Tessell directly or through the AWS or Azure marketplaces.
