Public, private or hybrid cloud? Make the smart choice!

You know you want to move to the cloud, but you don’t know which of the three major options – public, private and hybrid – are right for you. We’re here to help with this quick overview of the options, as well as the pros and cons of each.

Public Cloud

Think of this as a server in the sky. Public cloud, also known as infrastructure-as-a-service, provides the equivalent of a data center in a highly scalable, virtualized environment accessed over the internet. Customers can provision virtual servers – called “instances” – and pay only for the capacity they use. Many public cloud features are automated for self-service. Users can deploy their own servers when they wish and without IT’s involvement. Accounting and chargeback are automated. In fact, organizations often find the public cloud delivers the most significant savings not in equipment costs, but in administrative overhead.

The best applications to deploy in a public cloud are those that are already virtualized or that run on unmodified Linux or Windows operating systems. Commercial, off-the-shelf applications are a good example. Public cloud is also a good platform to use in evaluating and testing new applications, since many public cloud providers offer a wide variety of applications on a pay-as-you-go basis. Public cloud is also well suited to developing so-called “cloud native” applications, such as mobile apps.

Public cloud isn’t ideal for every use. Homegrown applications on legacy platforms or those with significant interdependencies may not migrate smoothly. Organizations that aren’t careful to manage instances can can end up paying for unused capacity. There are also hidden costs to be aware of, such as surcharges for data uploads and downloads or upcharges for guaranteed levels of service. Regulatory issues may also limit the use of public cloud for some applications entirely.Private Cloud

This is essentially a public cloud for use only by a single customer. Private clouds may be constructed on premises using virtualization and automation software, or licensed from service providers who deliver cloud services either from their own data centers or even on the customer’s own premises.

Private cloud is popular with companies that need tight control over data, whether for security, privacy or regulatory purposes. In regulated industries that specify how customer data must be stored and managed, it is sometimes the only cloud option. It’s also attractive for companies that need guaranteed service levels without the unpredictability of the public internet. Finally, private cloud provides the highest level of control for organizations that want deep visibility into who is using resources and how.

Private cloud is typically more expensive than public cloud because service providers must allocate capacity exclusively to the dedicated environment. However, that isn’t always the case. For companies with large capital investments in existing infrastructure, an on-premises private cloud is a good way to add flexibility, automation and self provisioning while preserving the value of their existing equipment. For predictable workloads, it can be the cheapest of the three models.

Hybrid Cloud

This is the most popular option for large corporations, and is expected to dominate the cloud landscape for the foreseeable future. Hybrid cloud combines elements of both public and private cloud in a way that enables organizations to shift workloads flexibly while keeping tight control over their most important assets. Companies typically move functions that are handled more efficiently to the public cloud but keep others in-house. The public cloud may act as an extension of an on-premises data center or be dedicated to specific uses, such as application development. For example, a mobile app developed in the public cloud may draw data from data stores in a private cloud.

Many of the benefits of hybrid cloud are the same as those of private cloud: control, security, privacy and guaranteed service levels. Organizations can keep their most sensitive data on premises but shift some of it to the public cloud at lower costs. They can also reduce costs by using public cloud to handle occasional spikes in activity that overtax their own infrastructure, a tactic known as “cloud bursting.” Hybrid cloud is also a transition stage that companies use as they move from on-premises to public cloud infrastructure.

There are many more dimensions to the public/private/hybrid cloud decision. A good managed service provider can help you understand the options and estimate the benefits and trade-offs.

Understanding and adopting Splunk

Splunk has been a trend in the industry for quite some time, but what do we know about its use and the market Splunk is targeting?

Splunk comes from the word “spelunking”, which refers to the activities of locating, exploring, studying and mapping.

  1. Data indexing: Splunk collects data from different locations, combines them and stores them in a centralized index.
  2. Using indexes for searches: the use of indexes gives Splunk a high degree of speed when searching for problem sources.
  3. Filtering results: Splunk provides user with several tools for filtering results, for faster detection of problems.

For more than a year I have been experimenting with Splunk in several facets: security, storage, infrastructure, telecom and more. We at ESI have a very complete laboratory which allowed me to push my experiments.

In addition to using all these amounts of data, I used open data to experiment with Splunk’s ability to interpret them.

I tested the open data of the site “montreal.bixi.com”; this is raw data formatted as follows:

Start date –  Start station number –  Start station –  End date –  End station number –  End station –  Account type – Total duration (ms)

With this data, we are able to find the most common routes, estimate the average duration of a trip, the anchorage points most requested for the entry or exit of bicycles.

For the operations team of the service, this provides real-time or predicted for the next day which anchors should have more bicycles, and mostly where these bicycles will go. They could predict the lack or surplus of bikes in the anchor points. If data is collected in real-time, alerts could be issued to indicate potential shortage or surplus in the anchor points. Thus the system facilitates planning and allows to be proactive to meet demand, rather than reactive. We would even be able to detect an undelivered bicycle; for instance a bike that has not been anchored for more than 24 hours could issue an alert, so the operations team attempts to trace it.

For marketers, one might think this data is useless, while the opposite is true; the same data can be used to put in place potential offers to attract customers, since we have the data that give the time of departure and arrival, time of use of the trips, and the most used routes. One can thus know the most used time slots and make promotions or adjust the rates according to objectives of traffic or customer loyalty.

For the management, open data unfortunately does not give the price of races according to the status of the users (members or non-members), but the beauty of Splunk is that one can enrich the collected data with data coming from a third-party system, a database or simply manually collected data. Management could then obtain reports and dashboards based on various factors, such as user status, travel time, days of the week, and much more. We could even make comparisons with previous months or the same month of the previous year. The applications are virtually limitless with data that resides in Splunk: the only limitation is that of our imagination!

These are of course fictitious examples made with available open data, but which could be real with your own systems and data.

The collection of information from a website can provide visibility for all users of a company, operations receive system overload alerts, marketers get information about the origin of the connections to target their campaigns based on this data, management gets a view of the user experience, as well as performance metrics that confirm SLAs.

Whether it is security, operations, marketing, analytics or whatever, Splunk can address your needs. In addition to the 1,200 applications available in its portal, you can create your own tables, reports, or alerts. You can use their Power Pivot to allow people to easily use the data and build their own dashboard.

The platform is easy to use and does not require special expertise: you only need the data there.

Do not hesitate to contact ESI for a presentation or a demo; it will be my pleasure to show you how to “Splunk”.

Guillaume Paré
Senior Consultant, Architecture & Technologies – ESI Technologies