I’ve got a couple of positive comments to my last note about hybrid clouds: “Is LATAM already prepared to face Hybrid Cloud Solutions?” that makes me pointing out some other thoughts to this kind of topics in regards to solutions known as multi-cloud managers.
This note addresses some of the trends that could make this open cloud model to get it absolutely commoditized to the point to getting it free of charge in the medium term, and pay for managed services and better security features (see my note “The difference between an open standard and an enterprise cloud model”)
Multi-Cloud Management has very interesting solutions for end-customer to optimize resources and financials.
Solutions like rightscale, enstratius or scalr bring a consolidated management for cloud resources (servers, block storage, object storage, network, etc.), templates, scripts/tools (chef, auto-scaling), events, alerts and monitoring among different global services providers and privates clouds based on known platforms like cloudstack and openstack.
Some of them can bring analytics like forecasting to plan future costs and get some advice to reduce them.
This online management applications are not addressed to cloud providers, actually is addressed to end-customers to take advantage from different global cloud offerings and other cloud options (privates or local cloud providers).
This applications make so much easier to go between public and private clouds speeding the hybrid model through just a couple of clicks. You can create the same conditions in both worlds and define rules that allow you to expand your resources from private to public and backwards.
Get hybrid through just a couple of clicks…
However, there are functionalities that IMHO must to be developed in a future time to make it more worthy to users that are not so expert.
- They must address platform. Actually, you can manage infrastructure, a “nice to have” is to manage platform. I mean, you can get some templates to build platform within minutes like wordpress. However, if they could be already built-in , be addressable from a common library at different sizes (see my suggestion about I/O profile analytics) and be spread into different cloud providers, this can be the easier way to save money: if they can really manage platform, they could swap or get more resources from different cloud providers transparently to the user in order to save more money, prices change between providers all the time.
- Real auto-scaling and high availability: When I mean they can manage platform, that could bring you a real auto-scaling feature between different could providers. They have to create referenced architectures that help you to get better availability through different HA Zones. Sites –like Asgar with Netflix and the concept “Design to Fail”- or even different Cloud Providers. If you need to provide auto-scaling and high availability to a only one web portal, you could get a built-in design solution that can work on top of two or three different providers, but this solution requires a special three layers design to expand and replicate even databases among these different public cloud options. Then, more than to use CPU, disk or memory usage thresholds to trigger some scaling actions, you should manage it based on the amount of user connections or user response timing.
- Security: It would be nice if they can identify security controls and make some suggestions about how to comply some part of the security standards like PCI, SOX, ISO27001.
- They have to bring some I/O Profile analytics to size applications easily. With all the information, you can manage these applications every day; they could make some big data analytics and bring some advice to size resources to the most known applications. For example, if I need to install a database, what is the best cloud and resources to get the best performance for a defined amount of user connections?
- Reference architectures to manage hybrid: Again, manage platform helps you to ease the construction of hybrid architectures. They should bring cents of options to build hybrid with built-in rules to move resources between clouds. Also, these options should take care by themselves of replications capabilities and migrate transparently to the end-user any server, storage or network resource based on performance or security rules.