CIOs are finding that they and their teams are increasingly and unavoidably living in a multicloud world. The question is: how well are they managing multicloud?
This was the question that I posed recently to the #CIOChat.
Biggest Management Challenges with Multicloud
CIOs suggest the challenges come in the following order:
1) the complexity of management
2) knowing where data and work loads are
3) cross platform vendor integration
4) acquiring security, privacy, and performance visibility from sub providers
5) reduced overall vendor leverage
Data, without question, is a major issue for CIOs to solve with Multicloud. Former CIO Tim McBreen says, “data has always been my biggest issue. Especially data that is needed across cloud platforms and requires constant synch. The same is true with cloud data warehouses and analytic data. Both require effort to make sure they work and are delivered at their contracted cost per unit of data moved and stored. This goes back to knowing what you are doing and maintaining and evolving your catalog, architecture, and strategy.”
Clearly, organizations should make a conscious choice about the data that should or should not be migrated into a cloud data warehouse. CIO David Seidl said in response to McBreen, “I wonder how many are doing much deep cross-vendor platform integration? Or if they’re running mostly distinct services based on underlying capabilities or requirements and then doing some higher level or API integration.”
Analyst Dan Kirsch said for this reason “everything needs to start by understanding where your data and workloads are. Multicloud, in most cases, was started by developers and business units. You now need to manage everything like a single platform — except it spans to multiple environments. Nevertheless, figuring out whose responsibility it is for managing IT assets is often a problem.
“For example, who owns data lakes in terms of data quality, security, etc. Meanwhile, some of an organizations most valuable data can remained locked in enterprise and SaaS applications plus of course on-premises databases.”
At the same time, CIO Martin Davis suggests that a “fragmented situation creates complexity and requires additional management and integrations. Good enterprise architecture is a must to avoid this.”
The worries of former CIO Ken LeBlanc go beyond data issues. He says, “there is the complexity of managing and viewing utilization and associated invoicing. All of this reduces buying power leverage.” Former CIO Joanna Young says “the management of multiple vendors and related contracts remains a difficult process. This includes the fitting the often-changing needs of a portfolio of components from different vendors.”
Clearly, CIO leadership is essential. For this reason, CIO Pedro Martinez Puig says, “CIOs need to set a clear choice path, so the decisions are consistent with all organizational stakeholders. It is important to keep a dual or multi provider during negotiations, plus not rely 100% on external expertise to monitor, plan and execute the projects and ongoing business. We can outsource everything but accountability.”
Is a Service Catalog Needed to Keep Track of Applications?
CIOs are amazingly candid about the current state of knowing where things are located. CIOs, nevertheless, believe strong central control is required to knowing all the places the organization has applications and infrastructure.
McBreen claims for example, “CIO should not be surprised at an outage to find out things aren’t working. They might uncover a nuance or something that was changed and not evolved. I go down to each cloud provider. I do not track their sub service providers outside of contractually making sure every provider including sub providers prove they meet certain security, privacy, and functionality requirements. We perform yearly audits as well.”
Seidl, however, was a bit hesitant to declare a service catalog is needed. “You need a way to keep track of what you’re running, what it’s costing, if it’s being used, if it’s scaled right, and if it’s being tested, reviewed, monitored, and managed. This is more – in my opinion – than a service catalog. And even if you do responsible cloud assessments for multicloud, how many layers deep do you go? What is the service provider’s service provider? Is there that does show up in your design or dependency map?” CIOs need to have:
- High level designs
- Real architecture diagrams and documentation
- Software/code defined structures for what is deploying in IaaS
- A mature deployment pipeline and process
- Useful monitoring, tracking, reporting, and alerting
Puig asserts for these reasons, “a catalog is needed including a model SaaS/PaaS/IaaS for each cloud contract, renegotiation points, when to choose what. And this requires a quite different skills to manage vendors and catalog effectively. You need a way to keep track of what you’re running, what it’s costing you, if it’s being used, if it’s scaled right, and if it’s being tested, reviewed, monitored, and managed.”
CIO Deb Gildersleeve agrees with Seidl and Puig when she says,“it is a really useful piece of information when you are assessing systems and tech debt.” And former CIO Joanna Young suggests that “configuration and asset management apply no matter where applications or workloads are located. CIOs must know where stuff is.”
Clearly, the need for a service catalog is essential to success, but it needs to be an automated tool as opposed to just a spreadsheet. Kirsch adds,“old best practices, like service catalogs and service-oriented architecture are critical in the cloud. Lifting and shifting poor practices will just move your problems to the cloud.”
Do Cloud Aggregators Make Management Easier?
Seidl claims, “cloud aggregators and multicloud platforms can help with data management. Cloud costs, however, are another story. Each cloud platform is architected differently. To decrease spend you need to be able to optimize workloads and data for the architecture.”
Young agrees and says that “it is situational. I would look at aggregators to see if effectiveness or efficiency advantages are being achieved for customers. Either way, you must know what services you need. Don’t underestimate the vendor/contract/services management whether aggregate or not.”
Meanwhile, Seidl suggests “what I am looking forward to is portable infrastructure that can land in the most cost-effective cloud based on a smart management engine while not causing disruptions. I don’t think we’re there in clean way yet.”
However, Puig claims “If your size does not permit to dedicate internal talent to manage it, aggregators can be a safe bet. But once you reach a critical size, you’d better keep a strong cloud management expertise internally to ensure you achieve desired cost savings.”
Challenges Multicloud Creates for Data Architecture
Young asserts that “data architecture represents a different kind of complexity. In fairness a lot of organizations with all or most of the computing on premises do not have this well-organized either. Ideally, IT organizations need to get this sorted before making a move to multicloud. The key message is whatever is a mess on-premises will be a mess in multi-cloud. Clean it up before or as part of switching and moving. Cloud in and of itself is not a cleansing agent instead it’s a different business model.”
Puig adds to Young’s remarks, “compliance risks multiply if you are working in the international arena.” For this reason, Seidl says that identity integration is great, but what about the data and business logic layer.” He goes onto say that “service duplication is a real challenge in SaaS. Everybody, for example, has a workflow engine.”
Kirsch concludes by saying “getting value from data in a multicloud environment is why business are looking at data virtualization and data fabric solutions. Having a consistent data management layer is essential. With this, you can connect to data wherever it resides including enterprise apps, SaaS, etc. versus moving data.”
Making Multicloud Easier to Manage
Young believes that “pragmatic and transparent services and related tools are needed that deliver upon the promises of simplification and efficiency. This way enterprises can focus their talent on differentiated technology.” Seidl adds that “with the small handful of major incumbents, this reminds me of other segments where up and comers have built compatible interfaces to allow easy adoption. Or built an abstraction layer to slice pennies out by doing things more transparently.”
Kirsch adds from the analyst corner that he is looking for “emerging technology to help with multicloud:
- Applying AI to Ops – AIOps
- Cloud Management across multiple cloud infrastructures to give visibility into costs, security and enforce governance rules
- Data fabric to analyze data no matter where it resides. Normalizing data from each cloud vendors proprietary management tools is a challenge to achieve across clouds.
I’ve seen businesses create their own multicloud management solutions using Hadoop. The problem is they need a team just to manage the system. I have seen businesses create their own multicloud management solutions using Hadoop. The problem is that they need a team just to manage the system.”
Gildersleeve agrees and suggests, “the big areas where process and technology can make multicloud easier is enhancing overall visibility and making data more accessible. Data in a silo is not helpful, so making it accessible is critical.”
Without question, CIOs have entered a multicloud era regardless of whether they planned for it. This means there is a need to make it easier to operate within the multicloud. At the same time, there is a huge need for data transparency and management. And this includes transparency around data compliance. The well governed are clearly going forward going to be the winners in the multicloud.