What to choose: PaaS and IaaS
In our blog on Habré, we like to sort out interesting cases related to the practical side of using virtual infrastructure by startups. In addition, we draw attention to foreign experience - we analyze everything related to the work of complex IT systems, infrastructure and hardware.
For example, recently we told:
Today we will tell about the main differences between PaaS and IaaS.
')
/ photo by Fondo Antiguo de la Biblioteca de la Universidad de Sevilla CC
The main difference is the level of abstraction over the equipment, which is somehow located in the data center. Each solution is suitable for the respective tasks.
What are we talking about:
Pros and cons of dedicated servers ("bare metal"):
Unfortunately, the disadvantages here often outweigh the possibilities, which require serious experience. Virtualization comes to the rescue when a single physical server shares several virtual servers. Technologies: Xen, KVM, VMware and Hyper-V and many more.
Advantages and disadvantages:
This means software access to computing power and storage, as well as network resources and configurations. You only need to pay for the use of data center resources.
Now, businesses do not need to think about buying their equipment and spend money on administration. You can very quickly launch pilot projects and quickly scale those that have become successful.
Automation of various parts of the data center has already happened: storage, network technologies and a huge number of other systems that required a certain set of skills were managed through the API, which opened up new horizons of opportunities for the developer community.
An example of a PaaS provider is Heroku. It runs completely on top of AWS. Without IaaS, platforms like Heroku most likely would not have appeared. PaaS providers use IaaS to provide their services.
PaaS allowed us to further reduce the time that passes from the moment the idea came to the deployment of the application - even more automation of the processes that had to be performed manually, as well as a reduction in the required level of knowledge to work with specialized systems.
Creating a PaaS is very difficult. Here you will need to create an application management database that monitors changes in Git, working versions and meta data of the application. In addition, cluster task scheduler, load balancer, automate DNS operation and implement a certain form of containerization (FreeBSD Jail, Solaris Zones, Linux Containers).
In simple terms, IaaS resembles leasing a server for a certain period of time. PaaS can be described as something more complex: a bundled set of tools to start and manage your web application.
Neighbors working on the same PaaS provider can affect each other's performance. In the case of IaaS, you can use larger instances, which most often means dedicated access to a physical machine that does not have “neighbors”. More PaaS system components mean more risk points where intervention may be required.
Although most PaaS platforms provide a universal interface for managing applications. Bribes the ease with which you can duplicate whole environments. You can run a clone of your entire system in just a couple of minutes. Here are just the cost of some platforms, for example, Heroku, can quickly become too large (even by the standards of IaaS).
It turned out that PaaS wins in some areas, and IaaS wins in others. That is the reality. If you can afford a greater degree of risk, then PaaS offers flexible application management capabilities that significantly speed up the process, compared to IaaS.
Additional reading:
For example, recently we told:
- how Spotify scales the Apache Storm ,
- considered deep learning hardware
- and talked briefly about an example of optimizing bandwidth on Ethernet networks .
Today we will tell about the main differences between PaaS and IaaS.
')
/ photo by Fondo Antiguo de la Biblioteca de la Universidad de Sevilla CCA bit of terminology
The main difference is the level of abstraction over the equipment, which is somehow located in the data center. Each solution is suitable for the respective tasks.
What are we talking about:
- Infrastructure as a service (IaaS);
- Platform as a service (PaaS).
Pros and cons of dedicated servers ("bare metal"):
- Performance - direct access to hardware, but also more difficult to manage;
- Reliability - but only if you yourself do not make a mistake in service;
- Resource use - when using dedicated servers, processes do not fight with other processes or virtual machines for the computational power of the processor, memory and network resources;
- Price - you pay only for the hardware (rent it from your provider).
Unfortunately, the disadvantages here often outweigh the possibilities, which require serious experience. Virtualization comes to the rescue when a single physical server shares several virtual servers. Technologies: Xen, KVM, VMware and Hyper-V and many more.
Advantages and disadvantages:
- Fast virtual machine cloning;
- Simple backups of virtual machine images;
- Working with virtual machines requires expertise.
Infrastructure as a service
This means software access to computing power and storage, as well as network resources and configurations. You only need to pay for the use of data center resources.
Now, businesses do not need to think about buying their equipment and spend money on administration. You can very quickly launch pilot projects and quickly scale those that have become successful.
Automation of various parts of the data center has already happened: storage, network technologies and a huge number of other systems that required a certain set of skills were managed through the API, which opened up new horizons of opportunities for the developer community.
Platform as a service
An example of a PaaS provider is Heroku. It runs completely on top of AWS. Without IaaS, platforms like Heroku most likely would not have appeared. PaaS providers use IaaS to provide their services.
PaaS allowed us to further reduce the time that passes from the moment the idea came to the deployment of the application - even more automation of the processes that had to be performed manually, as well as a reduction in the required level of knowledge to work with specialized systems.
Creating a PaaS is very difficult. Here you will need to create an application management database that monitors changes in Git, working versions and meta data of the application. In addition, cluster task scheduler, load balancer, automate DNS operation and implement a certain form of containerization (FreeBSD Jail, Solaris Zones, Linux Containers).
findings
In simple terms, IaaS resembles leasing a server for a certain period of time. PaaS can be described as something more complex: a bundled set of tools to start and manage your web application.
Neighbors working on the same PaaS provider can affect each other's performance. In the case of IaaS, you can use larger instances, which most often means dedicated access to a physical machine that does not have “neighbors”. More PaaS system components mean more risk points where intervention may be required.
Although most PaaS platforms provide a universal interface for managing applications. Bribes the ease with which you can duplicate whole environments. You can run a clone of your entire system in just a couple of minutes. Here are just the cost of some platforms, for example, Heroku, can quickly become too large (even by the standards of IaaS).
It turned out that PaaS wins in some areas, and IaaS wins in others. That is the reality. If you can afford a greater degree of risk, then PaaS offers flexible application management capabilities that significantly speed up the process, compared to IaaS.
Additional reading:
- How clouds develop: 6 statistical facts
- Everything in a new way: How moving to the cloud changes the role of internal IT service
- Why does business need a cloud: an overview of real IaaS usage scenarios
- How to evaluate the benefits of moving to the cloud
- SaaS, PaaS, IaaS, or your hardware: what do domestic IT companies use?
Source: https://habr.com/ru/post/275085/
All Articles