Wednesday, February 20, 2019

Cloud Technologies clouding our judgement ?

Now a days whenever a topic about storage comes up, first idea that comes to our mind is the cloud technologies. No, this article is NOT negative attitude about the cloud techs, but to remind us that there are still other options.

When I first came across the ideology of cloud technologies, roughly 10 years ago, I admit I was pessimistic. I mean, as an information security professional, first thoughts ever came to my simple mind was facts such as data security, encryption, availability etc. etc. But after 10 years, just like everyone else, I am more easy with those crucial questions as most services are protected and defined by international laws and legislations.  For instance the GDPR law in Europe (EU) 2016/679.

We like it or not, cloud technologies are already in our life. Our photos, mails, web servers and many other infrastructure applications are all about the cloud.

But everything should have to be. In recent years as the tech behind data storage and data transmission got more advanced, the TCO of hardware reduced, whilst, apart from the IT crowd, the corporations saw the values and advantages of hosting their data off-site.

Several advantages including:
  • Availability: With the right tools, your data can be accessible to you and the parties you allow to be shared with globally.
  • Reliability: As the data is distributed across several locations and servers, you know that your data is more safe than your local or networked storage and even can be bound by SLAs. 
  • Security: Several encryption and security options are widely available.
  • Performance: Of course as the data is located at a datacenter, the bandwidth requirements can be arranged by the service provider according to needs. Hence, most datacenters are on multiple load balanced data line backbones, allowing the desired throughput performance.

I am not going to mention about the Maintainability & Serviceability as you already know that as long as it is available, you can maintain and service your information according to your desire.

But apart from the key factors of Service Design, cloud hosts are trying to extend their range of their storage services to hosting big data. Just like 10 years ago unprecedented amount of articles are being posted online trying to persuade customers to believe that “big data” belongs to the cloud and it should also be hosted online, whilst some are even talking about backing up data to the cloud. 

This is the point I have to disagree, because big data means bigger responsibility. Don’t get me wrong, I am no stale mind and am always open to new perspectives. 
  • Though some would feel safe by placing their data on the cloud for reliability and availability when the topic is big data and backup:
  • The global data transfer rates are still limited which means more data throughput required, which means need for speed. 
  • Big data generally means bigger information or chunks on information, which means it is tougher to maintain and doing this remotely means it is tougher to service remotely (of course then you would require your server to be in the cloud too).

Assuring the reliability of the service remotely is a huge challenge both for service providers as well as customers as it is very difficult to fathom. Even if your service is bound by SLA’s, it’s no guarantee that the service will be delivered successfully. 

As I have observed at several cloud customers, whilst their data saved on the cloud, most of them still used cached copies on the local sites to be on the safe side or cache their active cloud data to the local constantly for backup purposes. Perhaps this is a good method but once again the data needs constant syncing to the cloud and if it’s database data we are talking about, it can mean a lot of work for the database people. Still, local caching clearly shows a sign of insecurity.

Coming back to our topic, big data doesn’t always mean “active” data. In most cases big data means incremental information that is added to the existing the pile. Perhaps the end of part of the data chunk may be active, but the beginning of the data may actually even be stale; just kept for regulatory or in case of need. 

So, carrying big data to the cloud actually is a purposeless idea. Especially for backup hosting ! 

Of course, the service providers do try to oppose otherwise, but when it comes to calculating the ROI and the TCO, the results still show that keeping the data on the local whilst backing up with traditional methods are not only financially convenient, but also still the most feasible method. 

Especially if your data is not just huge databases, but such as media data, for video sites etc., the availability and performance is utmost importance where ultimate performance is needed. Currently most sites who deal with audio visual sites still host their data on their own data centres or as we call it private clouds where they use their own methods to compress and throttle the throughput for optimum results. 

I mean, even the so called “out of date” magnetic tape backups are still more convenient way of backing up all sorts of data compared to the cloud hosts as the LTO tapes can hold 6.25 Tb data whilst soon to arrive LTO7’s are to be holding 25 Tb of data on a single tape where the data transfer rate is nearly at 1 Tb per second.

As a conclusion, OK, for apps and services that need to be globally available online from several locations, cloud services are advised, but for bigdata, the good old storage systems that are based at customer site are the most efficient and effective solution still.