DZone Research: The Most Important Elements of the Cloud
To learn more about what providers consider to be the most decisive elements of cloud computing, we asked over 30 industry executives and experts for their take.
Join the DZone community and get the full member experience.Join For Free
To gather insights on the current and future state of the cloud, we talked to IT executives from 33 companies about their, and their clients’, use of the cloud. We asked, "What do you see as the most important elements of the cloud?" Here's what they told us:
- On-demand provisioning and scalability.
- Business agility and flexibility. Customers can start up and scale applications as their businesses grow. However, economics may vary significantly, particularly when multi-cloud scenarios are considered.
- Innovation cycles the vendors are providing customers. The ability to rapidly scale up and down to meet customer demands geographically. How to get more velocity out of our team to reduce time to implement features. Focus on AWS.
- Focused on IoT using video IoT and other data to make the product superior to on-premise solutions. Business optimization as scale. Rapid deployment. Continuous updates. Leveraging latest AI/ML technologies.
- From our perspective, the most important elements include high-availability, scalability, and maintenance. Using the cloud, as an active-active architecture is much easier to implement than if an organization were to try to set up active-active using traditional deployment models. The cloud also makes bursting to scale much easier as well. Granted, an organization could deploy an infrastructure that could handle bursting to scale, but with the cloud, the resources and their management are much easier. For our customers, the maintenance of an enterprise application is completely negated because we do all the patching and upgrades for them.
- Easy machine availability -> Scalability, flexibility - getting computing resource quickly - for example, if I need a 16 CPU machine, I can just go into my resources and launch what I need immediately. Managed DB service -> scalability, automated backup tools. Server abstraction using serverless lambda functions and API Gateway.
- Simplicity and reliability – always on. AWS is like shopping on Amazon.
- We see the following four elements as the important elements of the cloud: 1) Pay per use, on demand. 2) Near unlimited availability. 3) Availability of specialized resources (GPUs, big memory etc.). 4) Smaller companies often don’t have the capacity for power, cooling or physical space for additional servers.
- 1) Availability 99.9999%. Can I trust a cloud provider? How do you build the confidence? 2) Security and privacy. Clarity on data ownership.
- When hosted in the cloud, you are no longer in control of your physical environment. Putting data there is fine to reduce labor cost and physical data server footprint. Security is now in someone else’s hands. Don’t go for the cheapest provider. Put a liability agreement in place.
- [Security is] someone else’s problem and you get geographic redundancy. Provided backups are encrypted, you have a good level of security. How to deal with large datasets. First full back up is copied to a hard drive and then shipped. Take data and move onto storage.
- The networking needs to be a foundational element. Get architecture correct to build off of. Get the landing zone correct. How to organize the account along with virtual private clouds. The value of automation as they move to the cloud. System-wide view of what’s going on. Have a bar for skillsets for the members of your cloud team. Public cloud is a different type of environment. Networking tends to be quite difficult. Good public cloud education. Five steps: 1) secure connectivity; 2) on-prem connectivity is agile; 3) automation; 4) segmentation; 5) visibility. Software-defined cloud routing allows for centralization and automation.
- Backup and security. AWS is to blame that the cloud is redundant. 2018 State of AWS Protection Report shows how they use AWS and how to protect data within it. Production environment and workload already in the cloud and 60% perform no backups and fewer than 10% have a recovery plan in place. Codespaces hacked in 2014. Deleted environment and out of business overnight. We protect critical client on AWS like Notre Dame, Harvard.
- The opportunity to convert traditional capital expenditures for infrastructure into operational expenditure, with higher flexibility and near-instant provisioning via APIs is extremely attractive to us. Aside from a couple racks of servers for esoteric platforms, we do not maintain or own any infrastructure. It is all in the cloud. This allows for predictability in cost control.
- Most customers are looking where public cloud fits. There are five stages that need to be evaluated continuously: 1) Figure out the associated costs, gap analysis on-prem versus public cloud. For on-prem consider peak time plus 2X or 3X. Understand how the application performs in the infrastructure. Use automation to burst. 2) Don’t do a proof of concept focus on production workload. Build a cross-functional team to move the workload to the cloud. Cloud is transformational and impacts more than the IT department. Cultural change requires everyone’s buy-in. Forces you to think about security up front. Create a foundation for success. 3) Evolution of the gap analysis. Learn skills needed on the team, and the process for moving from on-prem to cloud. Tools needed. Learn all applications and dependencies. If you don’t understand traffic patterns you won’t understand security and costs. Move groups for security, compliance and reduce costs. Not all apps can move to the cloud in the current state. Create an application roadmap on which apps can move and how to do. Forcing customers to buy holistically versus independently. Take an ecosystem approach to solving the problem. 4) You have a plan. How to execute. Avoid doing things manually. Understand infrastructure as code and DevOps. Extend operational guidelines into the cloud. Build scripts and templates. Build configuration management and deployment to remove human error. 5) Optimization. Ongoing real-time insights into usage, performance, and spend for end-user experience and to control costs. Hold lines of business accountable with chargebacks.
- Certain things are commodity don’t sweat the details. Memory is cheap. Ability to run an SQL server. Focus on cost of migration, support. How easy to transfer and migrate? If you are running a business with a lot of computation and analysis who has the best image processing systems? Look at differences based on business needs.
- In our view — and this certainly carries over to our own product design — simplicity, predictable and competitive pricing, and scale are the most important elements of the cloud. While some cloud providers are becoming more complex in what they offer, we’ve focused on keeping our products simple to use and flexible, so there is no lock in and developers can use things however they wish. Learning and documentation are also very important to developers building applications in the cloud and we do our best to serve them by making learning easy and accessible to everyone.
- The openness of the platform. Amazon is a very close P2. Both very open have taken different historical directions. Hosted VMs started hot and we’ve shifted toward containerization. Don’t be precious. Don’t get into a religion war. Between the main players around AI/ML if look at Google, IBM, Microsoft, AWS, Salesforce – what is the basis for their advantage in AI/ML/NLP. Once you get the body of data behind this it’s hard for someone to catch up. Think about providers for a particular workstream.
- I think it’s the ability to not only leverage others’ infrastructure but also their expertise. PaaS represents a significant shift in the supply and demand dynamic for enterprises. IT shops still need significant numbers of DevOps folks, but they can be deployed further up the stack.
- Best practices. No one asks for credentials. You get the keys with the credit card. Learn to do things properly to save time. Have a Center of Excellence around transformation journeys. Align skills and learning around roadmaps. Ascertain skills and how to get from A to B.
- Look at the problem from an application perspective. What’s the workload and the problem you’re trying to solve? From infrastructure centric to app-centric. What are the apps I’m trying to handle? Without an application lens, it’s hard to make the right choices between public and private cloud.
- The cloud is a part of the solution. Understand it’s not a binary decision. Processing is getting more distributed. Cloud, on-prem, edge computing. Cross-application and cross-analysis require a data fabric to be more powerful and easier. Customers acquiring companies with different public clouds. Data fabric enables data access across clouds. Other cloud implementation is too big to move.
- Tread carefully, tread slowly. You tell us where you want to work, and we’ll do our best to do what you want. We get nervous about recommending where to place the database.
- Toolsets to discover process and defensibility when data needs to be kept or destroyed. California privacy law will spur 49 other privacy laws. We do our best to honor the geography. Try to grab additional business context metadata for searching, discovering, and applying rules for privacy and governance.
- Today’s customers want instant fulfillment so the ease and speed of deployment the cloud offers are paramount to engaging customers on a trial basis first, so they can evaluate the effectiveness of the solution before committing to a paid service.
- Beneficial view, best practices, stages of maturity. Foundational themes 1) understand it’s not an ERP project, the cloud is a progression, a lot of virtualization in place start to cut cost with KVM or Open Stack and deploy into cloud-optimized 2) Shy customers away from complexity. Simplify cloud management. Move away from a model of paying millions to stand up a cloud – drive immediate business value. 3) DevOps teams that are cross-functional. Roles will vary. Be nudged in a direction to make DevOps model possible. Mode 2 operation creating new digital centers unshackling developers from the corporate stack. Think in a different way to create apps. 4) No organization is cloud-optimized. Not all applications are cloud optimized. Greenfield uses the cloud, but a big portion continues to live on VMs and containers. Legacy apps require refactoring but not reformatting.
- Experience is the same across the clouds so where ever the customer has their stack in a number of regions.
- The most important elements of the cloud are the SaaS players who provide high-quality data services that work around some surprising size and other limitations found in products like Azure CosmosDB and Google Datastore. For our own work, not to mention that for our clients, the most important element is knowing the limitations and pitfalls of the products you choose to work with.
Here’s who we talked to:
- Bikram Gupta Sarma, Vice President of Platform Engineering, Actian
- Glenn Luft, V.P. of Operations and Bill Tolson, V.P. of Marketing, Archive360
- Andreas Pettersson, CEO, Arcules
- Steven Mih, CEO and Frank Cabri, V.P. of Marketing, Aviatrix
- Linus Chang, Founder, BackupAssist and Scram Software
- Tim Jefferson, VP Public Cloud, Barracuda Networks
- Julian Dunn, Director of Product Marketing, Chef
- Andrew Larkin, Head of Content, Cloud Academy
- Sabapathy Arumugam, Co-founder and CTO, CoreStack
- OJ Ngo, CTO, DH2i
- Jeff Chou, CEO, Diamanti
- Shiven Ramji, VP, Product at DigitalOcean
- Kelly Stirman, CMO, Dremio
- Mike Milinkovich, Executive Director, Eclipse Foundation
- Lucas R. Vogel, Principal Consultant, Endpoint Systems
- Jason McGee, VP and CTO, Cloud Platform, IBM
- Erik Kaulberg, Senior Director, INFINIDAT
- Rich Petersen, President and Co-founder, JetStream Software
- Mark Brewer, CEO, Lightbend
- Adnan Mahmud, CEO and Founder, LiveStories
- Jack Norris, V.P. Data and Applications, and Bill Peterson, V.P. Industry Solutions, MapR
- Ryan Duguid, SVP Technology Strategy, Nintex
- Sebastian Straub, Senior Systems Engineer, N2WS
- Chris Brown, Technical Marketing Manager, Nutanix
- Vamsi Kiran Chemitiganti, Chief Strategist, Platform9
- Mike LaFleur, Global Director of Solution Architecture, Provenir
- Simon Galbraith, CEO, Red Gate Software
- Jamie Tischart, V.P. Operations, SendGrid
- Christian Kleinerman, Vice President, Product, Snowflake
- Aaron Brooks, Director of Innovation, Softchoice
- Rafi Rainshtein, Vice President of R&D and DevOps, SysAid
- Rob Lalonde, VP, GM NavOps, Univa
- Rajesh Ganesan, Vice President, Zoho Corporation
Opinions expressed by DZone contributors are their own.