I had an opportunity recently to learn a lot about Softlayer and get certified in Softlayer Solution Design, so I would like here to share my insights about some features, which may come as very useful for the post-digital world.
First of all I need to say that I work at IBM, and Softlayer is an IBM company. Softlayer cloud offering is actually, together with cognitive computing and Watson division, one of the strategic imperatives of the Big Blue. But I’m not directly attached to the Cloud Unit, so this certification was a challenge. IBM is known for the rigor in professional certifications. To get certified, it takes a lot of learning, practical exercises and experience, considerable time to prepare your certification package, and several levels of technical and peer reviews. All this hard work and learning has to fit into your regular work schedule, so in the end being certified at IBM is something IBMers are usually very proud of.
For those who know cloud, IBM has two cloud offerings – traditional Cloud Managed Services (CMS), with traditional stack: Power7 or 8 machines with AIX LPARs, SAN storage based on XIV systems, and on top of that old-fashioned middleware like DB2, Oracle, Websphere, SAP etc. The provisioning is semi-automatic, but you get fully managed, shared environment provided from 13 data centers in various world locations. This is quite good for legacy workloads.
Younger and more dynamic offering comes from Softlayer, originally a US company acquired in 2013. IBM has in meanwhile invested a lot in Softlayer brand, its global footprint, but kept its’ rather independent path to the market. Softlayer’s stack is similar to what you find in other cloud computing centers, but there are some important features that distinguish it substantially from its competitors.
But let me go a bit back to the basics. Cloud has been developing really very fast in the last years in terms of user adoption and breadth of services. You can’t imagine any mobile application nowadays without relying on cloud resources. And PaaS seems to be the standard. Companies have learned that platforms are the way how you disrupt the market in the most efficient way, so everyone wants to built his own and largest digital platform. Amazon AWS and Microsoft Azure are positioning themselves in this area of Platforms-as-a-Service. IBM Softlayer has focused on IaaS, which seems a bit out of fashion, but this actually allows many other companies to work on building their own platforms and disrupt the incumbent “disruptors”. And namely this Softlayer’s IaaS (Infrastructure as a Service) has the potential to bring the best value as the underlying technology for disruptive post-digital platforms. Here is why:
Global footprint
26 data centers around the world with presence in all continents, with exception of mainland China (Shanghai is expected to be available soon, in meanwhile Hong Kong can be used instead). In Europe, Softlayer has data centers in London, Amsterdam, Frankfurt, Paris, Milan and Oslo. Don’t cry that Softlayer is not present in Switzerland, for just keeping your data in Switzerland doesn’t guarantee any security. Your Swiss servers can also be infected with malware that can stay planted for months or years undetected, until the moment arrives (like in DNC leak recently). Learn how to patch and secure your systems, do the intrusion tests regularly, and if you are paranoid about NSA or similar entities, keep your most sensitive data on an isolated network in your cellar and encrypt the rest that you want to put outside.
All these 26 data centers are interconnected through a private and unmetered IBM network, which means that you get for free any traffic between your US, European and Asian systems. If you want to scale easily and cheaply your applications through different geographical markets, this comes very handy.
Flexibility
Autoscaling is very easy – define a group of VMs and create criteria based on which you want to add new VMs or shrink after a period of inactivity. You pay per hour, or if you want a discount, choose monthly payment scheme. Virtual servers are created or spawned in the matter of minutes and the process is fully automated.
Not only this – flexibility means being able to have your own dedicated environment with any software you want to bring along and install yourself. Order a bare metal server from the service catalog and after half an hour you have it running. You can request to add disks and use it as local and dedicated storage system (Softlayer recommends using Quantastor solution), or you can install for example a hypervisor and have your own dedicated private cloud based on OpenStack or VMware. The service is designed to accommodate any unique or complex design an organization might want to keep.
You can also combine these resources. Imagine for example having public facing Web servers on shared VMs, and keeping database servers on dedicated bare metal machines. But you can also move from one type of resource to another – use FlexImage to migrate if you want to experiment with a different architectures. This means flexibility.
Performance
Bare metals are faster than public VMs. And they are particularly faster when accelerated with NVIDIA GPUs. So you can add a Tesla to your server to make miracles possible. Not adding Serbian genius Tesla or best electric car Tesla, but this Tesla K80 with 5’000 cores that will make your deep learning or other artificial intelligence systems respond in fraction of a second. Basically with Softlayer you can build your own branded Tesla analytics solutions or Tesla recommender systems. And yes, GPUs are dedicated, not shared among customers, and chargeable on monthly basis.
Private network between data centers means consistent performance. If a company builds games or provide content, it doesn’t want the viewers to experience glitches or delays.
As for the public and shared VMs, Softlayer does not do overbooking. What you order is what you get, so you are sure to get consistent performance with your VMs.
Integration
You must have heard of Hybrid Cloud. It is a world where companies will decide to keep some operations at home (within their cellars) while some other workloads will be automated, public, and dropped somewhere else. And this makes sense – some data simply does not fit the public utility services model.
If you want to design and build such a system, you can easily integrate and manage it yourself with Softlayer management APIs. You can also load balance a combination of global and hybrid web resources with Softlayer load balancers like Citrix’ NetScaler.
Security
Softlayer segregates the traffic on three isolated networks: private for the traffic between your own servers, public for your internet connections, and management for Softalyer backups and monitoring. By simply disabling public ports on your VMs you exclude it completely from any attempt to connect to it from internet.
Apart from this, you have a range of options to further secure your environment: standard dedicated hardware firewalls to secure multiple servers in a single public VLAN, or dedicated Vyatta solution in case of complex architectures with multiple VLANs. Of course, the best practices would require that you put your first firewall to protect your publicly facing machines, and then segregate them from the private internal network with an additional firewall. But the system is designed to accommodate all variations of complex security architectures.
Another interesting feature – on Customer portal you can see exactly the location of all your public VMs – which POD, rack, position in the rack, piece of hardware. This allows complete transparency and better control.
Options
Imagine a disaster recovery solution where you don’t need the secondary data center. This is enabled with AltaVault, provided in partnership with NetApp. How does it work ? Simple: have Softlayer help you configure it in your private cloud, activate the solution on the ESX cluster in your primary data center , so that it replicates all images, application data and backups to your Softlayer account. In case of a disaster, and only then, you will order bare metal machines (you remembered that they arrive in 30 minutes ?), and do the restore of your VMs. It seems very easy and it can save a lot of money, still spent on expensive data replications and redundancy solutions. You pay for the machines only when you need to recover your operations , or test your DR procedures.
IBM Cloud Object Storage based on Cleversafe – this is very innovative. I’m still not very up-to-date with the details on internal workings, but this is a true alternative to OpenStack Swift. Object storage is used for internet – to store all static data like photos, texts, archives, web pages, catalog items, discussion forums, even videos. And Swift is terribly resilient – keeping three copies of data means that your storage is rock solid. But this doesn’t makes it very “green”, and Cleversafe avoids this over-redundancy problem by combining and splitting the data (in a similar way like another known IBM GPFS solution). Another benefit – Cloud Object Storage will be cheaper than Amazon S3. Expect to get it soon on Softlayer portal and for customers who already use Softlayer Swift-based storage, the migration path is very easy – access APIs will not change, so you can simply copy from one storage to another using un-metered private network. So, you will win both on performance and saving money.
Note to the readers: I know this overview is very technical. But technology is changing the world in front of our eyes, so even CEOs should get familiar with the ways how it works and what it could mean to their business. Neglect it at your peril. Many businesses wanted to remain “high level” or stuck to their outdated paradigms just to see themselves being gobbled up like Kodak or Nokia. For the business and IT professionals likewise: if you are not one year close to retirement, learn the new technology diligently – that is how you can provide many useful digital solutions for the modern world problems.
Conclusion
Softlayer is a promising solution for all businesses that want to grow digital and global. Contact me if you have questions on how it would fit into your business model, or if you consider similar technology options for your innovation roadmap.
Sasha Lazarevic
September 2016