By Cassandra Balentine
Organizations consider a multi-cloud strategy to handle business intelligence (BI) and corresponding data. The desire to avoid vendor lock-in while moving more business processes to the cloud makes this option attractive. However, there are challenges including added complexity.
Enterprises have increasingly implemented multi-cloud strategies over the last five to ten years, becoming more reliant on cloud service providers. However, each provider comes with its own rewards and challenges and organizations must weigh the pros and cons as it pertains to their own business. “While the cloud offers many opportunities for organizations, including greater flexibility and the ability to scale, there is also the risk of cloud lock in and being married to one vendor, which could cause difficulties down the line as newer, and potentially better technology solutions become available,” says Emma McGrattan, SVP of engineering, Actian.
Depending on an organization’s specific goals and processes, a multi-cloud approach to infrastructure is an option that should be considered and vetted as part of a BI strategy.
Evaluating a Multi-Cloud Strategy
First and foremost, a multi-cloud BI strategy should begin with a definitive goal. Once that has been established, enterprises should base their cloud partner evaluations on several factors, including the reputation and strengths of the potential cloud providers, and the expected complexity of migrating processes to the cloud.
“Focus first on the business outcomes you are trying to accomplish, and then look at the different technologies that will help you meet those goals,” suggests Roman Stanek, CEO/co-founder, GoodData. “By focusing on goals, you can then map to the different clouds as each offers something different. Enterprises also need to consider vendor lock in when mapping to the different goals and technologies.”
The primary considerations for a multi-cloud BI strategy are consistent with those for the overall data architecture. “These include managing costs, added resiliency, regulatory compliance with regard to data location, and reducing the risk of cloud vendor lock-in. It’s important to remember that multi-cloud doesn’t always give you automatic benefits, so you need to plan ahead on how to achieve an advantage. For example, if mitigation of cloud vendor lock-in is a main goal, then you need a strategy should you decide to part ways with a vendor. The exit strategy needs to incorporate plans on how to migrate to a different cloud site,” suggests Dale Kim, senior director, products/solutions, Arcadia Data.
Specific to BI, considerations include data access latency, user interface latency, and regulations around data by geography. “For data access latency, you should identify which data sets are related and how they can be co-located to minimize delays due to reading analysis across the network. For user interface latency, you should deploy the BI platform near the user to minimize the network impact of delivering outputs. And with certain regulations, you need to ensure that personal data does not leave a geographic boundary, so installing the BI platform in specific regions will help with compliance,” offers Kim.
McGrattan believes all organizations must become data driven and have the ability to handle any data with the right tool in any place to ensure superior business outcomes. “Businesses need a future-proof analytics solution that is designed to meet demanding enterprise requirements as they connect to existing and new data sources. In moving so much of their data and infrastructure to the cloud, companies must evaluate the performance, security, and deployment options available to them with a multi-cloud strategy and determine which solutions will give them the insights they need into their customers, markets, and operations to grow revenue and profitability while reducing risk.”
Path to Implementation
The path towards integration must be clear and goals well defined. Once the guidelines of integrating are set, specific BI components should be identified and addressed.
A multi-cloud strategy starts with planning and identification of the plan for gaining benefits. The strategy does not need to be defined prior to any cloud deployment, as the benefits can often be investigated after the initial cloud deployments. “Planning for BI should run alongside planning for general information and data management on multi-cloud. It is not a separate process, so all considerations should be made at the same time. For example, the data sources and the data scientists might be in one location, but the business users are elsewhere. It’s important to understand such characteristics so that if tradeoffs exist, they are known up front,” says Kim.
Trip Dixon, director of operations, iDashboards, says these considerations include existing and future silos, analytic needs, reporting requirements, dashboard needs, and data transfer and access requirements. Identifying which components should be implemented on each of the cloud providers will be the major design exercise. “Deploying each component to the cloud provider best positions to host that process is the goal.”
Stanek suggests starting with business outcomes and defining users and the various personas and use cases. “Once those have been agreed upon, map to the available data. That will give you the direction to take for the various clouds you will be using.”
McGrattan suggests organizations prepare for a multi-cloud by building their cloud-hosted applications for portability by taking advantage of technologies like Docker that abstract the operating system components. “IT departments need to get trained in the different security and management choices that different cloud platforms provide. The key is to align with vendors who also support a multi-cloud strategy. Having database or data integration technology available close to where an organization chose to store and process their data minimizes the costs associated with transferring data between clouds. Even on-premises can be thought of as a private cloud for organizations that need greater control of their most regulated data. This hybrid of private and public cloud is where an enterprise data will increasingly reside along with supporting data management services.”
For managing BI, a multi-cloud approach is attractive for many reasons, including reduced reliance on one vendor in terms of both vendor lock in and backup purposes.
The desire of many organizations not to get locked into one cloud vendor is driving the demand for multi-cloud strategies. “This allows for better service negotiating power as well as being able to take advantage of different cloud providers’ unique strengths. It is also desirable to have a second cloud provider for failover in case of system failure and potentially to adhere to data storage regulation requirements,” shares Dixon.
The increase in digital commerce and meeting customers’ needs in real time has also led to today’s enterprises needing easy, real-time access to their data to derive insights and make crucial business decisions at a moment’s notice. “Implementing a multi-cloud strategy helps give organizations the competitive advantage they need. The lowering management costs, low storage costs, high availability, and the inherent elasticity of compute resources make cloud platforms very attractive,” shares McGrattan.
Kim points out that the core trend around multi-cloud is the ongoing increase in public cloud use. This includes the emergence of pure Infrastructure- and Platform-as-a-Service solutions that enable organizations to build their own solutions versus having to go with a single vendor.
“Organizations are committed to increased deployments in the cloud and are finding new ways to take advantage of the choices in the market. Many early cloud implementations were of the lift-and-shift variety where gains around faster deployment and pay-as-you-go were the key objectives,” continues Kim. Now organizations are more experienced with basic cloud deployment and are more knowledgeable about the broader benefits of the cloud. As cloud vendors try to take advantage of the interest in cloud deployments, it behooves enterprises to explore what can be done by adopting multiple cloud vendors.
Advantages and Limitations
We’ve already discussed advantages of a multi-cloud strategy in terms of what’s driving its adoption for BI processes. However, there are additional benefits, as well as limitations to consider.
Additional advantages of a multi-cloud BI strategy include cost savings, performance optimization, a lowered risk of DDoS attacks, and improved reliability.
Other general advantages of multi-cloud apply to BI as well. “For example, cost optimization is relevant, as you can pick the most inexpensive cloud storage solution and the cheapest compute solution and mix the two. You can also spread the risk by reducing your dependence on any single vendor. Other factors, such as such as disaster recovery and data loss resilience are important as well,” says Kim.
A multi-cloud strategy for BI supports proximity to end users, shares Kim. This reduces the network latency by putting the BI installation near the target end users. This also helps with regulations that require data to reside in specific locations. Since the BI platform resides in the region as required by regulations, organizations are easily able to prove that they are in compliance.
“Another advantage is the justification of decentralized IT teams that are dedicated to specific regions and user groups retain control. This lets local IT teams run their own deployments and address the needs of their local end users, instead of trying to create enterprise-wide standards that may not be applicable or ideal for many user groups,” says Kim.
With a multi-cloud strategy, it is possible to locate data close to where it is processed and accessed. “If a single cloud provider doesn’t have a location advantageous for performance, a second cloud provider possibly would. In addition, many cloud providers offer advanced analytics capabilities, but do not perform well in other areas. Using multiple providers will allow an organization to take advantage of best-of-breed capabilities for numerous providers,” says Dixon.
Stanek agrees, adding that the closer you get to your data the better it is. “Different clouds contain different data sets so you need to be aware of the data needed to meet desired business outcomes,” he shares.
There are also challenges and limitations to consider.
Dixon says one of the largest potential limitations of a multi-cloud strategy is an organization’s inability to host their data in the cloud due to compliance and governance limitations. “For example, a healthcare organization may need to adhere to HIPPA and would not be allowed to store data offsite,” he offers. Another limitation is the complexity of managing multiple providers versus just one. “Besides the effort needed to maintain multiple maintenance and support relationships if problems arise in the overall system, finger pointing amongst providers could happen.”
McGrattan agrees, noting that keeping up with data governance and compliance across vendors can be challenging. It can be challenging to know where data resides, and enterprises must be compliance with HIPAA regulations, CJIS regulations, and more recently—GDPR. “Therefore, companies must vet vendors carefully and perform regular IT health assessments,” points out McGrattan.
A multi-cloud strategy creates a challenge of complexity. “Enterprises will need to manage multiple technologies while also being aware of vendor lock-in. There is a fine line between managing the two needs while also understanding what data is available and how and why it will be used,” says Stanek.
“While a multi-cloud strategy offers many benefits, some of the challenges organizations may face is dealing with multiple vendors,” adds McGrattan. This means multiple contracts and points of contact. If companies aren’t thoughtful in their approach, having multiple vendors can also lead to redundancies in services offered, so it’s important that each service has a specific use.
“A multi-cloud strategy for BI potentially has the downside of added complexity in managing separate installations. Though this can be an advantage if each installation is managed by a local IT team, centralized IT organizations will face higher administrative overhead,” adds Kim.
“Another challenge is the fact that you no longer have to proverbial ‘one throat to choke.’ Depending on your deployment architecture it may be harder to track down issues and get resolution across multiple vendors. With multi-cloud, you have less leverage for negotiating favorable terms and discounts since you are investing less in any single vendor,” shares Kim.
Complex and Complementary
Organizations should consider the benefits and limitations of a multi-cloud strategy when it comes to managing business data. No matter how many or which vendors are decided, it is important to set clear goals and define a path to implementation to reduce complexities. SW
Jan2019, Software Magazine