Discover PerformanceHP Software's community for IT leaders // October 2013
5 keys to combining cloud and analytics
Cloud computing and the power of big data analytics are both high priorities for enterprise IT. Avoid these major pitfalls before cementing a combined strategy.
Many organizations are testing the waters of big data–capable analytics. Their initial forays are frequently little more than a proof of concept—a cloud cluster spun up in relative haste—with a short-term intention. These organizations may also find themselves wondering what to do with these orphaned test systems after drawing that first rush of analysis.
Whether you’re trying to institutionalize such a nascent cloud deployment after the fact or making a proactive policy decision to avoid creating one in the first place, at some point you’ll have to decide how the cloud will factor into your big data analytics strategy. The decision isn’t necessarily limited to on-premise vs. cloud. For some, a managed cloud may offer the best of both worlds.
The winning combination of fast, flexible deployment and low capital expenditure has made cloud a very popular choice for many line-of-business applications. Analytics and business intelligence is no exception.
The IT consumerization trend means business stakeholders throughout your organization are feeling empowered to use software-as-a-service as they see fit. They often choose such cloud-based solutions because they can get quick results without additional budget allocation.
But the quick fix of shadow IT has downsides:
- Siloed data—without meaning to, these initiatives can isolate data
- Lack of governance—these systems make an end run around data policies that protect the organization from compliance failure and legal risk
Ultimately, the IT department will have to expend effort and resources to double back and undo these initiatives, bringing them into the fold. And they will have to make sure they are optimized for competitive advantage.
Factoring the cloud
If you’re not already actively planning your strategy for big data business intelligence, it’s looming in your very near future. Now’s the time to plan ahead. Will your organization be served best by big data analytics in the cloud or on premise? Here are five factors to consider:
1. Data in the cloud can be expensive to load—and to move.
Cloud deployments can be very useful for not yet “big” data, but once those data sets start to get truly big (10, 20, or more terabytes), it can become financially prohibitive to load or move that data elsewhere.
2. Usage costs for public cloud can add up quickly.
Cloud providers charge on a utilization basis—turning a fixed cost into a variable one. Trouble is, as data gets big, so do those variable costs. At some point, it will be more cost- and resource-effective to buy your own hardware and maintain it yourself. If your data is not in the cloud already, estimate your usage costs up front before you make a deployment decision.
3. Lack of liability/indemnity may be a blocker.
Most providers have come a long way in certifying their clouds to meet various government and industry regulation requirements, but many still will not take legal responsibility if something goes wrong. Depending on your industry, such lack of indemnity may be a deal breaker, especially for sensitive data.
4. Sometimes proximity matters.
For very large data loads, it makes sense to minimize latency and avoid moving data across a network. Consider the size of your data and the velocity (rate of change). If you’re loading terabytes of data per day or streaming data constantly (e.g., for scientific applications), deploy your BI system as close to your data as you can.
5. For some, cloud is the culture.
Many organizations today cultivate a culture of entrepreneurialism, and these companies tend to do everything in the cloud. If your CIO has made cloud agility a corporate value, then cloud will likely be the preferred choice.
Consider the cloud in the middle
The larger your data deployment, the more expensive a public cloud can become. However, managed service providers offer an alternative that can be the best of both worlds.
A managed cloud is a leased cloud under contract, usually for a term of one, two, or three years. The provider purchases the hardware, provisions the environment, and performs ongoing maintenance on your behalf. You pay a monthly fee, which essentially amortizes the infrastructure and maintenance costs for you over the length of the contract.
A huge advantage of a managed cloud is a service-level agreement, which is generally not available with traditional cloud providers.
A managed cloud makes sense when usage fees become inhibitive. You’ll save significantly over the usage fees charged by public cloud providers, but still avoid the hit of having to buy and maintain infrastructure yourself.
As an added bonus, some managed cloud providers offer data science services as an add-on. Data science skills are in high demand, so this can be a great value for organizations that haven’t yet built up adequate skills in the internal team.
Find out how you can transform your approach to analytics with the expert help of HP’s Big Data Discovery Experience.
For running Vertica on the Amazon cloud, there’s an AMI pre-configured for Community Edition and Enterprise Edition, and you can try Vertica Solo, an evaluation version of Vertica configured with a less-expensive Amazon instance type.
Diving into disruptive technology trends like cloud, mobile, and Big Data, HP’s CEO talks about moving not just IT, but the whole enterprise, into a new era.
Senior executives from HP Vertica and HP Autonomy explain the next steps in really understanding your customers and building a profitable relationship with them.
Dig into strategic trends with our new Discover Performance Weekly video series, and go backstage at events like RSA.
Gigaom and Cerner discuss real examples of how advanced analytics can transform healthcare.
Keep up with the latest innovations in next-generation analytics.
Insights into the power of human information.