By Cameron Bahar, Senior Vice President, Storage and Data Management, Oracle
As the world now knows, cloud computing – with its scale-up-and-down resources and pay-as-you-go model – is a flexible and cost-efficient way to deploy IT infrastructure, databases and applications, especially compared to footing the high fixed costs and complexity of running your own data center. Despite these obvious advantages, the architectures of Gen 1 clouds left much to be desired.
Take block storage, for example. As higher performance options for block storage emerged on cloud platforms, most clouds introduced new, disparate volume types, tiers, and pricing models. That trend among those providers still applies today as they continue to offer and expand a complex set of volume types and tiers that need cumbersome and careful planning and conversion processes to provision storage properly or to move between tiers. This is certainly not simple and is the antithesis to the value an elastic cloud service should provide.
We’ve learned from that trend and avoided this complexity with Oracle Cloud Infrastructure (OCI) Gen2 Block Volume service since its inception. We offer one flexible volume type with a simple slider to control its performance. Customers can configure and dynamically adjust volume performance on demand for existing or new volumes without any impact to their applications.
True to the promise of cloud, they can start anywhere on the performance slider and change it anytime and as often as required. There is no need to carefully study, understand, optimize, plan, and select performance tiers and prices, thus freeing our customers from that complexity. All application workloads get the same predictable and steady performance backed by a performance SLA that is offered only by Oracle; from read-heavy workloads such as streaming applications, write-heavy IOT applications, sensory and log data collection, to mixed read/write workloads typical of enterprise transactional database systems.
Retail customers, for example, would benefit greatly from such simple and flexible block storage infrastructure. During peak holiday seasons when site traffic and transactions surge, the OCI Block Volume service performance slider helps dynamically address these load spikes without impacting applications. This agility allows our customers to only pay for the performance they use when they need it.
Furthermore, when volumes are detached and not used, their cost can be automatically tuned down to a lower cost option for additional savings. And there is more to come, as Oracle is planning to add a performance-based autotune feature that will automatically adjust volume performance and cost as needed without human intervention.
Using the latest gear to maximize performance and efficiency
In tech, there is always a lot of talk about first mover advantage. But sometimes big benefits lie in entering a market later. OCI joined the cloud fray after large cloud rivals but because it did so, it could apply lessons learned from those attempts—and also Oracle’s own earlier cloud effort—and could deploy newer technology from the get-go.
OCI’s block and file storage systems, for example, were built using only NVMe SSDs (Solid State Drives) from day one. Its timing enabled OCI to skip a whole generation of slower spinning disk hard drives that still store important customer data at legacy cloud platforms. Other cloud providers have been adding SSDs as an option, but still field lots of hard drives for customer use. OCI Block Volume service today offers high performance NVMe SSD storage at prices that other providers typically offer for hard drive options.
Naturally, OCI does use hard drives for some storage offerings such as Object Storage where speed of access is not a top priority, but for critical live data most enterprises want fast access and OCI is NVMe SSD all the way.
Bringing enterprise DNA to the cloud
Unlike earlier cloud entrants, which built their respective infrastructure with small startup customers in mind, Oracle focused on large Fortune 1000-type companies because enterprises comprised the bulk of its existing customer base. The company knew it needed to first build a public cloud that suited the unique demands of organizations already running Oracle databases and applications in-house.
For example, one principal characteristic of object storage is its ability to offer high availability and durability via replication across multiple availability domains (ADs), also known as Zones. There is a challenging tradeoff between response time and consistency when designing such distributed systems. Some cloud object stores opted for “eventual consistency” and chose to sometimes return inconsistent data across ADs in order to be able to provide satisfactory response times. Enterprise customers are used to and require “strong consistency,” meaning each update is synchronously committed across ADs and read after write semantics are honored.
Our engineers took up this challenge and found a way to achieve both strong consistency and excellent response times and raised the bar in the industry. This in turn forced competitors to reconsider their designs and follow suit. Job one for OCI engineers here was to build a solution that reduces the impedance mismatch between on-prem enterprise systems and our cloud solutions.
Same cloud storage services run here, there, everywhere
Businesses also need to be able to run the same services on-premises as they do in the cloud. Thus, all the storage services offered in the OCI public regions are also available in a smaller form factor that can run in customer-controlled data centers – which is not something legacy cloud providers were built to do.
Especially for companies in financial services, healthcare, and other industries, the ability to keep key data under their own control remains a key requirement for regulatory and privacy reasons.
That flexibility is available to them via Oracle Dedicated Region Cloud@Customer (DRCC). DRCC includes all the services and capabilities that are available in OCI public regions. Since the same technology runs both DRCC and OCI public regions, DRCC customers are assured that should some workloads need to run in the public cloud, they can do so, while other data remains segregated.
Passing the savings on to customers
Given what I’ve said about OCI services, that they use the latest hardware and highly optimized software, you might think that they would not be price competitive with the older public clouds. But, in fact, our use of the most modern hardware and software written to wring the most out of available resources, also means the resulting services are extremely performant and that we can offer them at a surprisingly low price.
For example, OCI’s highly reliable ultra-high-performance block storage service offers a whopping 300K IOPs (input/output operations per second) per volume with sub-millisecond latencies at a small fraction of the price of rival clouds. High performance is prohibitively expensive on other clouds, but not on OCI. OCI block storage, depending on volume size and performance levels, can be 15- to 25-times less expensive than the most comparable block storage option offered by one public cloud giant, for example.
But this is not the case of a provider offering services at or below cost to gain ground. Our technology choices, system architecture, and distributed software optimizations allow us to be efficient and pass these savings on to our customers.
Cloud promises many advantages, flexibility and cost-efficiency dominant among them. But it also offers the promise of continuous innovation and the reduction of technical debt for businesses. Oracle Cloud Infrastructure has been a beneficiary of this continuous innovation, and as opposed to passing down technical debt, it is passing innovation along to its customers in the form of a lucrative technical inheritance as well as thought leadership in the cloud.