Hong Kong has just survived (yet again) another major typhoon without suffering from any damage to its data center, network infrastructure or power supply facilities. Given the relatively secured data center infrastructure in Hong Kong, should enterprises care about cloud-based disaster recovery at all?
As in other parts of the world, Hong Kong enterprises want zero disruption to their business, and an infrastructure that can support the "always-on" world. Financial services companies want faster and ever more seamless operations, governments want better and more reliable access to services for the public, while hospitals seek always-on and always available patient data and systems.
"Every company, no matter the size, needs disaster recovery. We live in an age that an hour of downtime is not an option," said Raymond Goh, regional technical manager for Veeam in Asia and Japan.
Commonly known as "DR", disaster recovery has been existed for some 20 to 30 years, with the key market players being storage vendors like EMC, IBM and Hitachi Data Systems, to name a few, said Ziv Kedem (pictured, middle), co-founder and CEO of Zerto, a US-based DR and business continuity software maker.
For the purpose of DR, one would typically replicate the applications to another array in another data center. When a disaster strikes, one can failover the applications to the DR site in order to bring up the applications again. As such, even if the whole or part of a data center is down, it will not affect the running of the business applications that it hosts, Kedem explained.
Securing from beyond natural disasters
On Monday and yesterday, Hong Kong was hit by its first major typhoon of the year, Typhoon Nida. Some 403 trees fell, two low-lying areas were flooded, and 500 flights were rescheduled. But typical of all typhoons that the city experienced, Typhoon Nida did not cause any casualty or major damage, let alone "disasters." Blessed with this, should Hong Kong enterprises pay attention to DR? Is such corporate data 'insurance' worth paying for?
"DR should be an integral part of a company's data loss avoidance strategy and has to cater to multiple scenarios -- not just limited to natural disasters. In reality, the top causes of data loss are due to 'everyday disasters' such as hardware or system malfunctions, human error, software corruption, computer viruses and malware," Goh said. In a recent IT Policy Compliance Group study, it was suggested that 75% of all data loss incidents were due to human error. "It would be a major concern if enterprises were not concerned about DR in Hong Kong or any part of the world."
"Think about the massive Tianjin explosion that occurred in May 2015. Accidents that happened in data centers are not necessarily a result of natural disasters," said Alex Li (pictured, right), founder and CEO, Bizconline. "As long as a company's productivity is dependent on its IT usage, any disruption to IT or the resulting data loss will incur losses for the company -- worse still, the closure of business. In fact, after the 911 tragedy, just less than two percent of the companies that lost their data managed to survive."
"DR is like an insurance that everyone and every enterprise need, it is just a matter of how much you buy," Li remarked.
Cloud DR versus traditional DR
How is cloud-based DR different from traditional DR? For existing infrastructure-as-a-service users, can they simply rely on their cloud service providers for data backup purposes?
In the past, only large enterprise organizations had the necessary budget or IT staff to afford traditional DR in the form of a private secondary site. Because the production workloads mostly run at the primary site, the secondary site is rarely used. "This drives the cost of traditional DR even higher compared to its value, making it difficult to sell an ROI to upper management," Goh said. "DRaaS (DR-as-a-service) solves these challenges by leveraging virtualization and replication technologies to create off-site replicas of virtual machines."
"Existing IaaS users can rely on their cloud service providers for data backup," Goh said. But they should demand more than just data backup. Goh reiterated that DR users should demand availability from service providers -- data availability with rightful recovery time and point objective (RTPO), which is measured in minutes, not hours or days.
"Not all IaaS provide data backup with strategy like 30 daily, 12 monthly and 7 yearly," said Li. (This data backup strategy typically requires an organization to keep 30 daily copies, 12 monthly copies and 7 yearly copies of its data.) "And not everyone can guarantee the usability of snapshots. There is no clear rule on the level of compensation from cloud service providers in case the historical data cannot be retrieved. As such, some enterprises would host a ‘third copy’ of their data on cloud for better protection."
The problems with traditional DR solutions, said Kedem, is that they would often create another obstruction layer, which is not desirable. "Using Zerto's cloud-based DR solution, organizations would require just a few minutes, at most an hour, to install the cloud DR system on two data centers. This allows enterprise users to conduct failover tests in a bubbled network," he said.
Data center modernization drives cloud DR adoption
Given the clear advantages of cloud DR over traditional DR, what are the key drivers for Hong Kong enterprises to adopt cloud DR or DR-as-a-service (DRaaS)?
The biggest drivers for DRaaS adoption are mainly due to organizations modernizing their data centers by leveraging virtualization and storage technologies.
"With the exploding data growth and demand for faster IT services, and no tolerance for data loss or service downtime, IT decision makers face huge pressure to deliver site redundancy with stringent service level agreements, and regulatory compliance in the shortest possible time-to-market. Extending the on-premise data center to the cloud data DR services has proved to be the fastest and most cost-effective way to achieve this."
DRaaS also provides instant access to resources that address many of the existing adoption challenges of DR, said Goh. According to Veeam's survey conducted in October last year, titled "The DRaaS opportunity", the APAC-based survey respondents indicated that the top three drivers for enterprise to consider DRaaS are lack of available personnel (83%), lack of DR experience (83%) and lack of DR site (71%). "DRaaS allows end users to focus on the replication activity, plan the different RPO values needed for applications, and define the DR strategy and DR solution using business metrics rather than IT needs."
Varied level of cloud DR adoption in HK
In Hong Kong, there is a disparity in different industry practices and the regulatory environment. This results in varied levels of interest in some sectors in exploring new IT trends, and a varied level of cloud DR adoption in other sectors.
"The financial sector adopts a very careful and prudent approach to cloud services due to corporate and regional compliance, while the Hong Kong Government has already built their own private cloud services which offer back office, shared services like email, web and file hosting, office applications to their statutory boards of Hong Kong government. And some of the commercial sectors, like the manufacturing and logistics industries, are already fully dependent on cloud DR for their site recovery operations," said Goh.
The ICAC, for example, has been deploying Bizconline's DRaaS since 2006, according to Li.
"Hong Kong enterprises are mainly adopting hybrid cloud services with selected on-premise core services replicating into a cloud-based data center leveraging on virtualized infrastructure," Goh said. As such, performing DR drill and testing can be limited to once or twice per year, limited by their infrastructure and resources even though enterprises are aware they need to do more frequent DR tests. Nonetheless, technology is already available. Leveraging on DR replication to perform automated, lights-off DR workloads testing with total isolation and no impact to production workloads and minimal human intervention are required. Full or partial DR testing can be performed with intelligent software-based and secured network tunneling.
"Like insurance, IT is being more heavily relied on by Hong Kong enterprises. If a DR site like DRaaS is affordable and reliable, why shouldn't enterprises deploy it as a life insurance?" Li concluded.