• Not to miss any critical alerts.
• Not to raise too many false alerts, annoying users and burying critical alerts under the heap.
• Deliver alerts on specific objects and parameters to specific people/groups at specific time frames.
• Analyze data over very short (minutes) or very long (weeks) intervals.
• Provide an effective way for users to quickly check alerts, understand the exact conditions of why they were raised, and update alert thresholds if needed.

• Customers and Vendors
• DST and SRC Areas
• Customers => Areas or Areas => Vendors
• Customers => Areas => Vendors
• Disconnect codes
• Customers => Disconnect codes
• Customers => Areas => Disconnect codes
• Products or Rate plans
• Equipment, Trunks, or IPs
• SRC/DST numbers
• Switch nodes
• SNMP stats

• Current calls
• Current connected calls
• Attempts per hour
• Connected calls per hour
• Minutes per hour
• Current capacity

• ACD
• PSC
• ASR
• ABR
• NER
• CPS

• PDD, TTR, TTC
• % of 487 codes
• Number of hunts
• % of LNP/MNP
• Media parameters
• % of transcoding

• Cost per hour
• Price per hour
• Profit per hour
• Profit per minute
• Profit per call
• Current balance

• concurrent (eg: Active calls), polled every minute and shown as lines on charts,
• per-window (eg: ACD), calculated over a window of recent calls every minute and shown as lines,
• per-hour (eg: Minutes per hour), shown as bars.

Some parameters can be only concurrent, like Active calls, others make sense only if they are calculated over a period of time, like Call attempts, Minutes, or Profit.

The first example illustrates that even though there is a direct correlation between concurrent calls ( red line ) and per-hour minutes ( green bars ), these parameters can never be compared directly (note left and right axes).

Most quality parameters in 5gVision, however, have both per-window and per-hour representations. The second chart shows ACD as a line and per-hour bars.

• Alerts can be raised or cleared every minute if necessary, providing very quick response time.
• On the other hand, concurrent alerts may require some delicate tuning in order to avoid repeated alerts for parameters that may be changing wildly.

• Alerts are raised or cleared at the beginning of an hour for the previous hour. This makes them less timely.
• Values are averaged over a long period of time, so it is less possible that the alert will be triggered by a quick, but short variation of a parameter.

• the  red area  shows the interval from the tentative alert raise till sending the raise notification, this interval is called the assurance interval,
• the  green area  shows the interval from the tentative alert clear till sending the clear notification.

• the  red bar  indicates the hour for which the alert was raised,
• the  green bar  shows the hour in which the alert was cleared.

• First of all, it is configurable, you may have 0 delay, and all alert notifications will be sent right away, however:
• the delay may be needed to make sure the alert was not triggered by a quick variation of a parameter, and its value indeed went, and stayed below/above the threshold.
• In the picture to the left the ACD is going below 3 min. several times for a short period, but the alert is raised only when ACD stays low for more than 5 min.
• Same is true when alerts are cleared. We need to make sure the value not only went above/below the clearance threshold for a moment, but stayed at this level for some time.

The notification/assurance delay is only needed for concurrent stats. Per-hour alerts will always trigger notifications right away, at the beginning of each hour for the previously calculated hour.

• Absolute alerts (or ABS alerts) will compare a current parameter value to a threshold. This is good if you have a lot of objects of the same type with similar quality requirements. For instance, you sell 50 areas and you need an alert if ACD for any of them goes below 5 mins.
• But what if these 50 areas have different levels of good and bad ACD? Setting up 50 alerts with 50 different thresholds may be messy.
• Differential alerts (or DIFF alerts) will not care about the absolute value of a parameter, but will compare the previous and current values of a parameter, and trigger alerts if the drop/raise is below/above the allowed %% threshold.
• Thus, all your 50 areas may have different absolute ACD values, but may be covered by only one alert set on ACD going down more than 30%. Over which period of time? This is discussed next.

• Concurrent and per-window stats may change quickly minute by minute, so it makes sense to average the values over some period of time.
• 5gVision uses 10 minutes by default, but this can be changed. Consequently, the “current” value is the one averaged over the last 10 minutes.
• The “previous” value is averaged for the 10-minute interval from 20 to 30 minutes ago by default. The chart on the right demonstrates this.
• DIFF alerts for line-type stats can be raised from once a minute to once an hour.

• In case of Per-hour stats the hourly value of a parameter for the hour that has just completed is compared to the same parameter for the previous hour.

• If you see 2 adjacent  yellow  bars – they represent the “previous” and the “current” hours that were compared. If you see the  red  and  green  bars – this is the ABS alert first raised, then cleared.

• The concept of clearing an alert for the ABS method relies on a simple logic: the alert is cleared when the value goes over a certain “clear” threshold.
• An object, configured with an ABS alert, goes through 3 stages: the pre-alert (OK) condition, the alert stage, and the post-alert (OK) condition again.
• While an object is in the ABS alert stage, no repeated notifications are sent until the alert clears and a clear notification is dispatched.

• For DIFF alerts the clearing logic may not be so simple. If an alert is raised when ACD drops 30% over the last 30 min, it does not mean that ACD will eventually jumps 30% again, and we may capture this event. ACD may grow steadily over 3 hours, and this alert will never be cleared with the DIFF approach.
• DIFF alerts are raised on every occasion. If ACD drops from 9 to 6, and in 5 minutes drops further from 6 to 4 – there will be 2 alerts and 2 notifications on each drop.

	ABSOLUTE alerts (ABS)	DIFFERENTIAL alerts (DIFF)
C o n c u r r e n t	Raised and cleared every minute. Current, this minute values, are compared to thresholds. Notifications may be delayed to assure the alert/clear was not triggered by a quick variation of a parameter.	Raised (never cleared) every 5 minutes. Average values for the last 10 minutes are compared to average values within the interval of 20 to 30 minutes ago. Notifications are sent right away.
P e r - h o u r	Raised and cleared at the beginning of every hour for the previous hour. The values for the previous full hour are compared to thresholds. Notifications are sent on alert raise or clear and are always sent right away.	Raised (never cleared) at the beginning of every hour for the previous 2 hours. The values for the previous full hour are compared to the values of the hour before the previous one. Notifications are sent on alert raise only and are sent right away.

• An alert log will show every raised alert, even if email or SMS notifications are off.
• The alert history cell is a timeline with  red  bars for ABS or  yellow  bars for DIFF alerts and gives a quick idea of when each alert was raised and cleared over the last 2 hours. One bar represents 5 minutes.

• When the mouse is over the history bar, it will show times of alert raising and clearing.
• ABS alerts values/thresholds are highlighted in  red  for raises and  green  for clears.
• DIFF alerts values/thresholds are highlighted in  yellow .
• Double-clicking a row will yeild a chart for this object combination and a parameter.

• You may set up the minimum and maximum number of calls the object should have in order to be considered for alerts. For instance, an ACD alert for your major vendors may only be triggered if you are sending at least 50 active calls or 1000 calls per hour to a Vendor.
• In case of DIFF alerts, you may also set the minimum and maximum values for the parameter itself. Lets say that you have a DIFF alert on equipment for ACD drop over 40%. At the same time, you are not interested in GWs having ACD less than 0.5 min, as they are probably your test GWs, or the GWs with ACD over 10 min, since if it drops by 40% – it is still an acceptable ACD level.

While standard per-hour alerts are raised on the last 1-2 hours of stats, and DIFF alerts on concurrent or per-window stats are raised on stats for the last 30 minutes, custom intervals allow you to compare any interval in the past to any other interval or sum up/average stats over several hours and compare them to a threshold.

You may thus compare a parameter in the previous hour to the same hour yesterday or one week ago (on the picture). Or compare average minutes over the last 24 hours to average minutes of the previous week. Or, on the contrary, compare very short intervals, like active calls this minutes to the previous minute.

There are several ways to limit emails/SMSes delivered to users:

• Maximum number of SMSes sent at once. Configured per user. SMS messages are limited to 160 symbols, as a result, 5gVision may need to split the notification into several messages. This way, too many messages can arrive at once. The default system limit is 3 messages.
• Maximum number of Emails/SMSes sent per hour and per day. Configured per user.
• Maximum number of Emails/SMSes sent per month. Configured per user group. You may want to limit monthly SMSes to control costs.
• Each user group may have a different schedule when the emails or SMSes are allowed.
• Alerts can be easily switched off completely or in part if needed, for instance, when you are doing maintenance.