|Before introducing SIOS Coati||After introducing SIOS Coati|
|Built own monitoring system with Zabbix and Nagios||SIOS Coati supports auto-scale monitoring and configuration change is not required|
|If performing monitoring 24/7, it is entrusted to MSP and it costs much (more than 150,000 yen)||30,000 yen/month, reducing costs by about 80%|
|Receive failure notifications day and night and perform troubleshooting||Since Coati performs restoration, only you have to do is confirmation when the system recovers successfully|
"Monitoring is necessary for stable operation of web servers, but when using a monitoring tool, operators receive failure notifications but have to do subsequent work. There should be many infrastructure engineers who had been waken up in the middle of the night by the notification and took care of failure.
However, INFOBAHN's IT person said that he could go to sleep without worrying about such a thing. He talked about the secret."
You realize problems of the company's server operations only after the sytem is down
The number of companies that independently launch owned media and blogs for the purpose of sharing information on their products and related technologies is increasing. On the other hand, monitoring is necessary for stable operation of web servers, but if you outsource the operations to an external managed service provider (MSP), the cost shall increase more than expected. However, if you use a system monitoring tool such as Zabbix or Nagios, failure is notified but subsequent work needs to be done by operators. This is the same even if the monitoring is performed by MSP. There should be many infrastructure engineers who had been waken up in the middle of the night by the notification to take care of failure.
Yu Amano, IT Division Manager at INFOBAHN Group, Inc. who is responsible for infrastructure management, is one of them.
Infobahn is a digital agency that supports overall digital branding, including UX strategy, building and management of owned media and and is a subsidiary company of INFOBAHN Group, Inc. to which Mr. Amano belongs. In IT Division of INFOBAHN Group, Inc., designing and operation management of servers for customers' owned media at INFOBAHN, technical support for servers of Mediagene Inc. (running content media sites including "Gizmodo Japan," "cafeglove," "Life Hacker (Japanese edition)," "BUSINESS INSIDER JAPAN," "ROOMIE," etc.) and design, development, operation and management of internal servers.
IT Division has several staffs including Mr. Amano; however, many of them are mainly responsible for internal IT support or management of cloud-based groupware such as "G Suite," and operation and monitoring of servers had been performed by a small number of members including Mr. Amano.
Approximately 80% of customer servers and servers of the company's own media used Amazon Web Services (AWS), implemented CMS (Contents Management System) etc. and outsourced operations and monitoring to MSP. For customer projects, they can secure outsourcing expenses for monitoring from the project budget. For the company's media, the budget is also secured because there are sponsored content (such as smart news), which may directly lead to loss of opportunities when the site goes down, and it does not lose the planned accesses by introducing articles from external services.
However, in terms of the internal servers, "monitoring at that time was not enough, because monitoring was totally outsourced to an MSP and we did not consider the operations as "monitoring": the redundancy of the web servers and preventive detection capability were not enough." said Mr. Amano, looking back the time when he joined the company three years ago.
Infrastructure engineers cannot sleep well
only with a monitoring tool or MSP
However, failure notifications are sent day and night. Everytime he receives a notification, Amano has to get up and start restoration work. Daisuke Yoshioka, Executive Manager of Cloud Services Business Planning Dept. at SIOS Technology analyzes the background of issues that many enterprizes are facing.
"In the public cloud, users need to manage setting changes of operation monitoring tools when configuration is changed or instances are increased. Moreover, in case of trouble or when receiving a notification of performance deterioration, users have to take care of them day and night and therefore the increase of system management workloads has become an issue. Monitoring tools only send failure notifications and MSPs in charge of monitoring also need to handle them manually until recovery. Although the operation of AWS is simple and scalable, it tends to be overlooked that a required level of monitoring know-how, skills, resources also goes up accordingly. "
Under such circumstances, Amano came across SIOS Technology's "SIOS Coati." His company already have a relationship with SIOS through sales support of G Suite, and Coati's beta version was recommended when he told the company's situation.
"We performed verification many times including stopping the service, and confirmed that it started as expected. We could see it was reliable, and also found that automatic monitoring and restoration can be realized only with the initial simple configuration," said Amano.
Regarding the price, you can use Coati for free up to 10,000 hours for the first month. The pricing was pay-per-use, 3 yen per hour (tax excluded) for each instance (now 5 yen per hour for each instance). It was a satisfactory price range for them.
"We tend to consider that we can outsource monitoring of servers for customer systems but we have to monitor internal servers by ourselves. Even if we can do it by ourselves for the time being, it will eventually out of hand as the business expands. But when we decide to outsource monitoring, it costs more than 150,000 yen for a month and this is not cheap," said Amano.
It tends to be overlooked but monitoring requires lot of workloads and it is noticed only when costs are presented by MSPs. But by using SIOS Coati, it is no doubt that workloads and costs for monitoring can be reduced.
15-minute setting, smooth restoration as if no failure occurred
SIOS Coati is a cloud service that supports automatic monitoring and restoration of AWS EC2. It peers Amazon VPCs of Amazon VPC and SIOS, monitors them via AWS API and detects and recovers failure quickly. Because it performs monitoring in almost real-time and can correspond to scale-out, all the instances can be in scope of monitoring even when the number of the instances increases. It also provide detailed failure report by email so that IT personnel can reduce work on failure analysis.
It was in early March 2017 that the company started monitoring of the internal servers using SIOS Coati. "After submitting the setting information of peering at the time of online signing up, all I had to do was to configure the servers to monitor. Procedure manual for Windows and Linux configuration was provided and it took 10~15 minutes to complete the configuration by just clicking on the screen as described in the manual," said Mr. Amano.
Coati restores the system automatically and restarts the OS of the virtual machine when the restart of the service fails.Then it issues a failure report which covers from failure occurrence to successful recovery based on the relevant information of the virtual machine, services and AWS obtained before and after the recovery.
While Amano said that there had been no failure because it had not been long since the operation had started, "sometimes other IT members unintentionally made AD server down, and I received email notification from Coati when I was out saying that the service had been restored. Restoration was so smooth that I didn't even notice it until I was told," Mr. Amano laughed.
Amano says that he can relax because basically the service is automatically restored. "I no longer have to be waken up by failure notifications in the middle of the night to take care of them. It's nice to have automated and simplified processes from failure recovery to normal operation, even though for the internal servers. I will not be disturbed my sleep any more," said Amano.
SIOS Coati can be used widely including monitoring of WordPress and Mastodon
Currently we use Coati for some internal servers, but Amano said that he wants to monitor all servers managed by IT Division on AWS with SIOS Coati.
He also told us that he would like to use Coati for customer projects in the future.
"We would like to install redundant configuration for media site launch, even with the minimum configuration because we cannot secure sufficient budget. We can use a hosting service, but it could be expensive depending on services. But if we have SIOS Coati, we can install necessary CMS such as WordPress and Movable Type on one server with AWS and provide added value that the services recover automatically when it fails,” said Amano.
In that sense, in order to use it more widely, he expects SIOS Coati to expand functions such as automatic launch while watching processes, or automatic startup of virtual machines. He wants this especially because functional development can be expected.
"It is our new service, but because it is a cloud service, I'd like to receive such feedback directly and continue to enhance its functions more and more based on customers' demands," said Yoshioka. He also said that further enhancement including GUI enhancement to enable non-infrastructure engineers to use it easily, improvement of a failure report so that it can replace the internal failure report, and interaction with other monitoring systems by providing API. “It is important for us to consider how we can add values with automation while reducing costs; not think about what to offer based on customers’ budget.”
Stable operation is enabled with SIOS Coati for trending Mastodon servers. At first, Mastodon seemed to be no good because the service stability depended on the skill and work of server operators, but the number of companies that build their own social media on it is increasing. SIOS Coati can surely reduce the number of workloads and anxiety of such server operations.