Last week I had a briefing with ProfitBricks to get a better understanding of the infrastructure-as-a-service (IaaS) offering from Germany and to ask specific questions. I examined ProfitBricks critically with two articles in the past, number one here and number two here. First, because I don’t like marketing phrases that just promise much more than really is behind it and secondly because even the technical promises must be kept.
ProfitBricks presents itself as a typical infrastructure-as-a-service providers. A graphical web interface is used to form customized servers to an entire data center. A complex network structure shall ensure a real isolation of the customer’s network in the virtual datacenter. A meshed network with a redundant Infiniband connection and a highly redundant storage (including automatic backup) provides the performance and availability of data.
In addition, ProfitBricks has its own security team and highly experienced system administrators that provide a round the clock support.
ProfitBricks has data centers or co-locations in Germany (Karlsruhe) and the USA (Las Vegas). However, both data centers are not connected with each other – physical nor virtual. So no data exchange between Germany and the United States can take place in this way.
The infrastructure components
At ProfitBricks servers can be equip from 1 to 48 cores, and between 1 GB and 196 GB of RAM. The maximum is currently at 60 cores and 250 GB of RAM per server, which can be enabled through a support contact. Storage is available between 1 GB and 5000 GB. However, this can only be assigned to one server directly. So there is no central storage. To realize this, one has to build its own central storage appliance and distribute the storage over it.
Within a self-designed data center (see below), two zones (similar to Amazon Availability Zones) are available. This allows to configure two servers, whereby one of them will not notice about problems in the area of the second server.
There is no centralized firewall. Instead, each of all NICs in a server can be configured with their own rules. A central firewall can achieved by creating a dedicated firewall appliance (Linux + IPTables or a prepared commercial firewall as an ISO image).
Although a load balancer is available, ProfitBricks recommends to build an own one based on an appliance because among others the ProfitBricks one has no monitoring included.
ProfitBricks do not offer additional own value-added services. And on their own admission, this will never happen. Instead, the provider relies on a network of partners that will provide appropriate services for the infrastructure platform.
Currently unique: The Data Center Designer
What really convinced me at ProfitBricks is the “Data Center Design (DCD)”. Such a one no IaaS provider worldwide has it in this format.
Based on these graphical web interface one is able to create a complete virtual data center individually and activate the configuration or modify it with a mouse click – whether it’s about servers, storage, load balancers, firewalls, or the corresponding network.
Is a data center design ready, it can be saved and deployed. Previously the user obtained more information about a check through the system. Here is determined whether everything is configured correctly – e.g. that all servers have a boot drive with the corresponding image. Then the total cost per month for this virtual data center is itemized.
However, the DCD has still a weak point. Is a data center deployed, no single server can be removed from the design or stopped via the web interface. Therefore the entire data center must be un-deployed. Then the server is removed and the data center must be re-deployed. Using the proprietary SOAP API that supports, among other things Java and C #, a single server should be removed. This web feature is to follow, as well as a REST API, in the future.
The customer is mostly left on its own
ProfitBricks offers a German support, which has either worked for years as an administrator, or was involved in the development of the system. Support is included for free. Even if one evaluates the platform only with a test account.
Otherwise ProfitBricks is a common self-service like all the other IaaS providers. This means that a customer is self-responsible for the design of its virtual infrastructure and how an application on the infrastructure can scale and provided highly available.
With additional questions and approaches, for example for configuring a separate firewall appliance or a dedicated load balancer, external partners should help.
The billing follows exact to the minute per hour. The cost break down on this as follows:
- 1 Core = 0,04 EUR per hour
- (Windows Server plus 0,01 EUR per hour)
- 1 GB RAM = 0,09 EUR per hour
- 1 GB storage = 0,09 EUR per 30 days
- 1 GB traffic = 0,06 EUR per GB traffic
For the US market:
- 1 Core = 0,05 USD per hour
- (Windows Server plus 0,02 USD per hour)
- 1 GB RAM = 0,015 USD per hour
- 1 GB storage = 0,09 USD per 30 days
- 1 GB traffic = 0,08 USD per GB traffic
Live Vertical Scaling
ProfitBricks supports the so-called Live Vertical Scaling. This means that further resources such as CPU and RAM can be add to a virtual server during operation. This feature must be enabled separately for each server, and the server must then be restarted once.
However, what I have noted here and what has be confirmed by ProfitBricks during the briefing, the operating system, database, software and the own application must support it. The systems need to realize that suddenly more cores and RAM are available and use it. And in the opposite case also deal with, when the resources scale down again.
ProfitBricks is interesting
ProfitBricks is an interesting infrastructure-as-a-service offering. Especially in the very cloud-sparse (IaaS) Germany with a data center in Germany. Particularly noteworthy is the Data Center Designer (the only USP), which is currently unique in the world and provides convenience features, that the other IaaS providers are neglecting. Admittedly the designer rakes at one point or another (Example: removing server), but that will certainly change in a next version.
At the end of the day ProfitBricks is a pure infrastructure-as-a-service provider that has its strengths in infrastructure operations. This also showed the briefing. Therefore an interview with CEO Achim Weiss confuses me, which I had read a few weeks ago. Besides enterprises he called Internet startups as ProfitBricks target customers. Today, I consider this as an utopia. Without a services portfolio like the offering of the Amazon Web Services, this target group can not be achieved. The service gap can and should be closed by service partners. Another but quite legitimate approach if the strengths are in a different area.